Release v0.4.1 (#4686)

Co-authored-by: Reinier van der Leer <github@pwuts.nl>
Co-authored-by: Nicholas Tindle <nick@ntindle.com>
Co-authored-by: Nicholas Tindle <nicktindle@outlook.com>
Co-authored-by: k-boikov <64261260+k-boikov@users.noreply.github.com>
Co-authored-by: merwanehamadi <merwanehamadi@gmail.com>
Co-authored-by: Merwane Hamadi <merwanehamadi@gmail.com>
Co-authored-by: Richard Beales <rich@richbeales.net>
Co-authored-by: Luke K <2609441+lc0rp@users.noreply.github.com>
Co-authored-by: Luke K (pr-0f3t) <2609441+lc0rp@users.noreply.github.com>
Co-authored-by: Erik Peterson <e@eriklp.com>
Co-authored-by: Auto-GPT-Bot <github-bot@agpt.co>
Co-authored-by: Benny van der Lans <49377421+bfalans@users.noreply.github.com>
Co-authored-by: Jan <jan-github@phobia.de>
Co-authored-by: Robin Richtsfeld <robin.richtsfeld@gmail.com>
Co-authored-by: Marc Bornträger <marc.borntraeger@gmail.com>
Co-authored-by: Stefan Ayala <stefanayala3266@gmail.com>
Co-authored-by: javableu <45064273+javableu@users.noreply.github.com>
Co-authored-by: DGdev91 <DGdev91@users.noreply.github.com>
Co-authored-by: Kinance <kinance@gmail.com>
Co-authored-by: digger yu <digger-yu@outlook.com>
Co-authored-by: David <scenaristeur@gmail.com>
Co-authored-by: gravelBridge <john.tian31@gmail.com>
Fix Python CI "update cassettes" step (#4591)
fix CI (#4596)
Fix inverted logic for deny_command (#4563)
fix current_score.json generation (#4601)
Fix duckduckgo rate limiting (#4592)
Fix debug code challenge (#4632)
Fix issues with information retrieval challenge a (#4622)
fix issues with env configuration and .env.template (#4630)
Fix prompt issue causing 'No Command' issues and challenge to fail (#4623)
Fix benchmark logs (#4653)
Fix typo in docs/setup.md (#4613)
Fix run.sh shebang (#4561)
Fix autogpt docker image not working because missing prompt_settings (#4680)
Fix execute_command coming from plugins (#4730)

.devcontainer/devcontainer.json

@@ -7,11 +7,12 @@
     "ghcr.io/devcontainers/features/common-utils:2": {
       "installZsh": "true",
       "username": "vscode",
-      "userUid": "6942",
-      "userGid": "6942",
+      "userUid": "1000",
+      "userGid": "1000",
       "upgradePackages": "true"
     },
     "ghcr.io/devcontainers/features/desktop-lite:1": {},
+    "ghcr.io/devcontainers/features/github-cli:1": {},
     "ghcr.io/devcontainers/features/python:1": "none",
     "ghcr.io/devcontainers/features/node:1": "none",
     "ghcr.io/devcontainers/features/git:1": {
@@ -25,8 +26,20 @@
     "vscode": {
       // Set *default* container specific settings.json values on container create.
       "settings": {
-        "python.defaultInterpreterPath": "/usr/local/bin/python"
-      }
+        "python.defaultInterpreterPath": "/usr/local/bin/python",
+        "python.testing.pytestEnabled": true,
+        "python.testing.unittestEnabled": false
+      },
+      "extensions": [
+        "ms-python.python",
+        "VisualStudioExptTeam.vscodeintellicode",
+        "ms-python.vscode-pylance",
+        "ms-python.black-formatter",
+        "ms-python.isort",
+        "GitHub.vscode-pull-request-github",
+        "GitHub.copilot",
+        "github.vscode-github-actions"
+      ]
     }
   },
   // Use 'forwardPorts' to make a list of ports inside the container available locally.
@@ -36,5 +49,8 @@
   // "postCreateCommand": "pip3 install --user -r requirements.txt",
 
   // Set `remoteUser` to `root` to connect as root instead. More info: https://aka.ms/vscode-remote/containers/non-root.
-  "remoteUser": "vscode"
-}
+  "remoteUser": "vscode",
+
+  // Add the freshly containerized repo to the list of safe repositories
+  "postCreateCommand": "git config --global --add safe.directory /workspace/Auto-GPT && pip3 install --user -r requirements.txt"
+}

.devcontainer/docker-compose.yml

@@ -4,16 +4,9 @@ version: '3.9'
 
 services:
   auto-gpt:
-    depends_on:
-      - redis
     build:
       dockerfile: .devcontainer/Dockerfile
       context: ../
     tty: true
-    environment:
-      MEMORY_BACKEND: ${MEMORY_BACKEND:-redis}
-      REDIS_HOST: ${REDIS_HOST:-redis}
     volumes:
       - ../:/workspace/Auto-GPT
-  redis:
-    image: 'redis/redis-stack-server:latest'

.dockerignore

@@ -2,6 +2,7 @@
 *.template
 *.yaml
 *.yml
+!prompt_settings.yaml
 
 *.md
 *.png

.env.template

@@ -1,10 +1,16 @@
+# For further descriptions of these settings see docs/configuration/options.md or go to docs.agpt.co
+
 ################################################################################
 ### AUTO-GPT - GENERAL SETTINGS
 ################################################################################
 
+## OPENAI_API_KEY - OpenAI API Key (Example: my-openai-api-key)
+OPENAI_API_KEY=your-openai-api-key
+
 ## EXECUTE_LOCAL_COMMANDS - Allow local command execution (Default: False)
-## RESTRICT_TO_WORKSPACE - Restrict file operations to workspace ./auto_gpt_workspace (Default: True)
 # EXECUTE_LOCAL_COMMANDS=False
+
+## RESTRICT_TO_WORKSPACE - Restrict file operations to workspace ./auto_gpt_workspace (Default: True)
 # RESTRICT_TO_WORKSPACE=True
 
 ## USER_AGENT - Define the user-agent used by the requests library to browse website (string)
@@ -13,202 +19,186 @@
 ## AI_SETTINGS_FILE - Specifies which AI Settings file to use (defaults to ai_settings.yaml)
 # AI_SETTINGS_FILE=ai_settings.yaml
 
+## PLUGINS_CONFIG_FILE - The path to the plugins_config.yaml file (Default plugins_config.yaml)
+# PLUGINS_CONFIG_FILE=plugins_config.yaml
+
+## PROMPT_SETTINGS_FILE - Specifies which Prompt Settings file to use (defaults to prompt_settings.yaml)
+# PROMPT_SETTINGS_FILE=prompt_settings.yaml
+
+## OPENAI_API_BASE_URL - Custom url for the OpenAI API, useful for connecting to custom backends. No effect if USE_AZURE is true, leave blank to keep the default url 
+# the following is an example:
+# OPENAI_API_BASE_URL=http://localhost:443/v1
+
 ## AUTHORISE COMMAND KEY - Key to authorise commands
 # AUTHORISE_COMMAND_KEY=y
+
 ## EXIT_KEY - Key to exit AUTO-GPT
 # EXIT_KEY=n
 
+## PLAIN_OUTPUT - Plain output, which disables the spinner (Default: False)
+# PLAIN_OUTPUT=False
+
+## DISABLED_COMMAND_CATEGORIES - The list of categories of commands that are disabled (Default: None)
+# DISABLED_COMMAND_CATEGORIES=
+
 ################################################################################
 ### LLM PROVIDER
 ################################################################################
 
-### OPENAI
-## OPENAI_API_KEY - OpenAI API Key (Example: my-openai-api-key)
 ## TEMPERATURE - Sets temperature in OpenAI (Default: 0)
-## USE_AZURE - Use Azure OpenAI or not (Default: False)
-OPENAI_API_KEY=your-openai-api-key
 # TEMPERATURE=0
-# USE_AZURE=False
 
-### AZURE
-# moved to `azure.yaml.template`
+## OPENAI_ORGANIZATION - Your OpenAI Organization key (Default: None)
+# OPENAI_ORGANIZATION=
+
+## USE_AZURE - Use Azure OpenAI or not (Default: False)
+# USE_AZURE=False
 
 ################################################################################
 ### LLM MODELS
 ################################################################################
 
-## SMART_LLM_MODEL - Smart language model (Default: gpt-4)
+## SMART_LLM_MODEL - Smart language model (Default: gpt-3.5-turbo)
+# SMART_LLM_MODEL=gpt-3.5-turbo
+
 ## FAST_LLM_MODEL - Fast language model (Default: gpt-3.5-turbo)
-# SMART_LLM_MODEL=gpt-4
 # FAST_LLM_MODEL=gpt-3.5-turbo
 
-### LLM MODEL SETTINGS
-## FAST_TOKEN_LIMIT - Fast token limit for OpenAI (Default: 4000)
-## SMART_TOKEN_LIMIT - Smart token limit for OpenAI (Default: 8000)
-## When using --gpt3only this needs to be set to 4000.
-# FAST_TOKEN_LIMIT=4000
-# SMART_TOKEN_LIMIT=8000
+## EMBEDDING_MODEL - Model to use for creating embeddings
+# EMBEDDING_MODEL=text-embedding-ada-002
+
+################################################################################
+### SHELL EXECUTION
+################################################################################
+
+## SHELL_COMMAND_CONTROL - Whether to use "allowlist" or "denylist" to determine what shell commands can be executed (Default: denylist)
+# SHELL_COMMAND_CONTROL=denylist
+
+## ONLY if SHELL_COMMAND_CONTROL is set to denylist:
+## SHELL_DENYLIST - List of shell commands that ARE NOT allowed to be executed by Auto-GPT (Default: sudo,su)
+# SHELL_DENYLIST=sudo,su
+
+## ONLY if SHELL_COMMAND_CONTROL is set to allowlist:
+## SHELL_ALLOWLIST - List of shell commands that ARE allowed to be executed by Auto-GPT (Default: None)
+# SHELL_ALLOWLIST=
 
 ################################################################################
 ### MEMORY
 ################################################################################
 
-### MEMORY_BACKEND - Memory backend type
-## local - Default
-## pinecone - Pinecone (if configured)
-## redis - Redis (if configured)
-## milvus - Milvus (if configured - also works with Zilliz)
-## MEMORY_INDEX - Name of index created in Memory backend (Default: auto-gpt)
-# MEMORY_BACKEND=local
+### General
+
+## MEMORY_BACKEND - Memory backend type
+# MEMORY_BACKEND=json_file
+
+## MEMORY_INDEX - Value used in the Memory backend for scoping, naming, or indexing (Default: auto-gpt)
 # MEMORY_INDEX=auto-gpt
 
-### PINECONE
-## PINECONE_API_KEY - Pinecone API Key (Example: my-pinecone-api-key)
-## PINECONE_ENV - Pinecone environment (region) (Example: us-west-2)
-# PINECONE_API_KEY=your-pinecone-api-key
-# PINECONE_ENV=your-pinecone-region
+### Redis
 
-### REDIS
 ## REDIS_HOST - Redis host (Default: localhost, use "redis" for docker-compose)
-## REDIS_PORT - Redis port (Default: 6379)
-## REDIS_PASSWORD - Redis password (Default: "")
-## WIPE_REDIS_ON_START - Wipes data / index on start (Default: True)
 # REDIS_HOST=localhost
+
+## REDIS_PORT - Redis port (Default: 6379)
 # REDIS_PORT=6379
+
+## REDIS_PASSWORD - Redis password (Default: "")
 # REDIS_PASSWORD=
+
+## WIPE_REDIS_ON_START - Wipes data / index on start (Default: True)
 # WIPE_REDIS_ON_START=True
 
-### WEAVIATE
-## MEMORY_BACKEND - Use 'weaviate' to use Weaviate vector storage
-## WEAVIATE_HOST - Weaviate host IP
-## WEAVIATE_PORT - Weaviate host port
-## WEAVIATE_PROTOCOL - Weaviate host protocol (e.g. 'http')
-## USE_WEAVIATE_EMBEDDED - Whether to use Embedded Weaviate
-## WEAVIATE_EMBEDDED_PATH - File system path were to persist data when running Embedded Weaviate
-## WEAVIATE_USERNAME - Weaviate username
-## WEAVIATE_PASSWORD - Weaviate password
-## WEAVIATE_API_KEY - Weaviate API key if using API-key-based authentication
-# WEAVIATE_HOST="127.0.0.1"
-# WEAVIATE_PORT=8080
-# WEAVIATE_PROTOCOL="http"
-# USE_WEAVIATE_EMBEDDED=False
-# WEAVIATE_EMBEDDED_PATH="/home/me/.local/share/weaviate"
-# WEAVIATE_USERNAME=
-# WEAVIATE_PASSWORD=
-# WEAVIATE_API_KEY=
-
-### MILVUS
-## MILVUS_ADDR - Milvus remote address (e.g. localhost:19530, https://xxx-xxxx.xxxx.xxxx.zillizcloud.com:443)
-## MILVUS_USERNAME - username for your Milvus database
-## MILVUS_PASSWORD - password for your Milvus database
-## MILVUS_SECURE - True to enable TLS. (Default: False)
-##   Setting MILVUS_ADDR to a `https://` URL will override this setting.
-## MILVUS_COLLECTION - Milvus collection, change it if you want to start a new memory and retain the old memory.
-# MILVUS_ADDR=localhost:19530
-# MILVUS_USERNAME=
-# MILVUS_PASSWORD=
-# MILVUS_SECURE=
-# MILVUS_COLLECTION=autogpt
-
 ################################################################################
 ### IMAGE GENERATION PROVIDER
 ################################################################################
 
-### OPEN AI
-## IMAGE_PROVIDER - Image provider (Example: dalle)
-## IMAGE_SIZE - Image size (Example: 256)
-##   DALLE: 256, 512, 1024
+### Common
+
+## IMAGE_PROVIDER - Image provider (Default: dalle)
 # IMAGE_PROVIDER=dalle
+
+## IMAGE_SIZE - Image size (Default: 256)
 # IMAGE_SIZE=256
 
-### HUGGINGFACE
-## HUGGINGFACE_IMAGE_MODEL - Text-to-image model from Huggingface (Default: CompVis/stable-diffusion-v1-4)
-## HUGGINGFACE_API_TOKEN - HuggingFace API token (Example: my-huggingface-api-token)
-# HUGGINGFACE_IMAGE_MODEL=CompVis/stable-diffusion-v1-4
-# HUGGINGFACE_API_TOKEN=your-huggingface-api-token
+### Huggingface (IMAGE_PROVIDER=huggingface)
 
-### STABLE DIFFUSION WEBUI
-## SD_WEBUI_AUTH - Stable diffusion webui username:password pair (Example: username:password)
-## SD_WEBUI_URL - Stable diffusion webui API URL (Example: http://127.0.0.1:7860)
+## HUGGINGFACE_IMAGE_MODEL - Text-to-image model from Huggingface (Default: CompVis/stable-diffusion-v1-4)
+# HUGGINGFACE_IMAGE_MODEL=CompVis/stable-diffusion-v1-4
+
+## HUGGINGFACE_API_TOKEN - HuggingFace API token (Default: None)
+# HUGGINGFACE_API_TOKEN=
+
+### Stable Diffusion (IMAGE_PROVIDER=sdwebui)
+
+## SD_WEBUI_AUTH - Stable Diffusion Web UI username:password pair (Default: None)
 # SD_WEBUI_AUTH=
-# SD_WEBUI_URL=http://127.0.0.1:7860
+
+## SD_WEBUI_URL - Stable Diffusion Web UI API URL (Default: http://localhost:7860)
+# SD_WEBUI_URL=http://localhost:7860
 
 ################################################################################
 ### AUDIO TO TEXT PROVIDER
 ################################################################################
 
-### HUGGINGFACE
-# HUGGINGFACE_AUDIO_TO_TEXT_MODEL=facebook/wav2vec2-base-960h
+## AUDIO_TO_TEXT_PROVIDER - Audio-to-text provider (Default: huggingface)
+# AUDIO_TO_TEXT_PROVIDER=huggingface
+
+## HUGGINGFACE_AUDIO_TO_TEXT_MODEL - The model for HuggingFace to use (Default: CompVis/stable-diffusion-v1-4)
+# HUGGINGFACE_AUDIO_TO_TEXT_MODEL=CompVis/stable-diffusion-v1-4
 
 ################################################################################
-### GIT Provider for repository actions
-################################################################################
-
 ### GITHUB
-## GITHUB_API_KEY - Github API key / PAT (Example: github_pat_123)
-## GITHUB_USERNAME - Github username
-# GITHUB_API_KEY=github_pat_123
-# GITHUB_USERNAME=your-github-username
+################################################################################
+
+## GITHUB_API_KEY - Github API key / PAT (Default: None)
+# GITHUB_API_KEY=
+
+## GITHUB_USERNAME - Github username (Default: None)
+# GITHUB_USERNAME=
 
 ################################################################################
 ### WEB BROWSING
 ################################################################################
 
-### BROWSER
 ## HEADLESS_BROWSER - Whether to run the browser in headless mode (default: True)
-## USE_WEB_BROWSER - Sets the web-browser driver to use with selenium (default: chrome).
-##   Note: set this to either 'chrome', 'firefox', or 'safari' depending on your current browser
 # HEADLESS_BROWSER=True
+
+## USE_WEB_BROWSER - Sets the web-browser driver to use with selenium (default: chrome)
 # USE_WEB_BROWSER=chrome
-## BROWSE_CHUNK_MAX_LENGTH - When browsing website, define the length of chunks to summarize (in number of tokens, excluding the response. 75 % of FAST_TOKEN_LIMIT is usually wise )
+
+## BROWSE_CHUNK_MAX_LENGTH - When browsing website, define the length of chunks to summarize (Default: 3000)
 # BROWSE_CHUNK_MAX_LENGTH=3000
-## BROWSE_SPACY_LANGUAGE_MODEL is used to split sentences. Install additional languages via pip, and set the model name here. Example Chinese:  python -m spacy download zh_core_web_sm
+
+## BROWSE_SPACY_LANGUAGE_MODEL - spaCy language model](https://spacy.io/usage/models) to use when creating chunks. (Default: en_core_web_sm)
 # BROWSE_SPACY_LANGUAGE_MODEL=en_core_web_sm
 
-### GOOGLE
-## GOOGLE_API_KEY - Google API key (Example: my-google-api-key)
-## CUSTOM_SEARCH_ENGINE_ID - Custom search engine ID (Example: my-custom-search-engine-id)
-# GOOGLE_API_KEY=your-google-api-key
-# CUSTOM_SEARCH_ENGINE_ID=your-custom-search-engine-id
+## GOOGLE_API_KEY - Google API key (Default: None)
+# GOOGLE_API_KEY=
+
+## GOOGLE_CUSTOM_SEARCH_ENGINE_ID - Google custom search engine ID (Default: None)
+# GOOGLE_CUSTOM_SEARCH_ENGINE_ID=
 
 ################################################################################
-### TTS PROVIDER
+### TEXT TO SPEECH PROVIDER
 ################################################################################
 
-### MAC OS
-## USE_MAC_OS_TTS - Use Mac OS TTS or not (Default: False)
-# USE_MAC_OS_TTS=False
+## TEXT_TO_SPEECH_PROVIDER - Which Text to Speech provider to use (Default: gtts)
+# TEXT_TO_SPEECH_PROVIDER=gtts
 
-### STREAMELEMENTS
-## USE_BRIAN_TTS - Use Brian TTS or not (Default: False)
-# USE_BRIAN_TTS=False
+### Only if TEXT_TO_SPEECH_PROVIDER=streamelements
+## STREAMELEMENTS_VOICE - Voice to use for StreamElements (Default: Brian)
+# STREAMELEMENTS_VOICE=Brian
 
-### ELEVENLABS
-## ELEVENLABS_API_KEY - Eleven Labs API key (Example: my-elevenlabs-api-key)
-## ELEVENLABS_VOICE_1_ID - Eleven Labs voice 1 ID (Example: my-voice-id-1)
-## ELEVENLABS_VOICE_2_ID - Eleven Labs voice 2 ID (Example: my-voice-id-2)
-# ELEVENLABS_API_KEY=your-elevenlabs-api-key
-# ELEVENLABS_VOICE_1_ID=your-voice-id-1
-# ELEVENLABS_VOICE_2_ID=your-voice-id-2
+### Only if TEXT_TO_SPEECH_PROVIDER=elevenlabs
+## ELEVENLABS_API_KEY - Eleven Labs API key (Default: None)
+# ELEVENLABS_API_KEY=
+
+## ELEVENLABS_VOICE_ID - Eleven Labs voice ID (Example: None)
+# ELEVENLABS_VOICE_ID=
 
 ################################################################################
-### TWITTER API
+### CHAT MESSAGES
 ################################################################################
 
-# TW_CONSUMER_KEY=
-# TW_CONSUMER_SECRET=
-# TW_ACCESS_TOKEN=
-# TW_ACCESS_TOKEN_SECRET=
-
-################################################################################
-### ALLOWLISTED PLUGINS
-################################################################################
-
-#ALLOWLISTED_PLUGINS - Sets the listed plugins that are allowed (Example: plugin1,plugin2,plugin3)
-ALLOWLISTED_PLUGINS=
-
-################################################################################
-### CHAT PLUGIN SETTINGS
-################################################################################
-# CHAT_MESSAGES_ENABLED - Enable chat messages (Default: False)
+## CHAT_MESSAGES_ENABLED - Enable chat messages (Default: False)
 # CHAT_MESSAGES_ENABLED=False

.gitattributes

@@ -1,5 +1,5 @@
 # Exclude VCR cassettes from stats
-tests/**/cassettes/**.y*ml linguist-generated
+tests/Auto-GPT-test-cassettes/**/**.y*ml linguist-generated
 
 # Mark documentation as such
 docs/**.md linguist-documentation

.github/CODEOWNERS

@@ -0,0 +1 @@
+.github/workflows/ @Significant-Gravitas/Auto-GPT-Source

.github/ISSUE_TEMPLATE/1.bug.yml

@@ -8,14 +8,16 @@ body:
         ### ⚠️ Before you continue
         * Check out our [backlog], [roadmap] and join our [discord] to discuss what's going on
         * If you need help, you can ask in the [discussions] section or in [#tech-support]
-        * **Throughly search the [existing issues] before creating a new one**
-
+        * **Thoroughly search the [existing issues] before creating a new one**
+        * Read our [wiki page on Contributing]
         [backlog]: https://github.com/orgs/Significant-Gravitas/projects/1
         [roadmap]: https://github.com/orgs/Significant-Gravitas/projects/2  
         [discord]: https://discord.gg/autogpt
         [discussions]: https://github.com/Significant-Gravitas/Auto-GPT/discussions
         [#tech-support]: https://discord.com/channels/1092243196446249134/1092275629602394184
         [existing issues]: https://github.com/Significant-Gravitas/Auto-GPT/issues?q=is%3Aissue
+        [wiki page on Contributing]: https://github.com/Significant-Gravitas/Nexus/wiki/Contributing
+
   - type: checkboxes
     attributes:
       label: ⚠️ Search for existing issues first ⚠️
@@ -25,23 +27,29 @@ body:
       options:
         - label: I have searched the existing issues, and there is no existing issue for my problem
           required: true
+
   - type: markdown
     attributes:
       value: |
-        Please provide a searchable summary of the issue in the title above ⬆️.
-
-        ⚠️ SUPER-busy repo, please help the volunteer maintainers.
-        The less time we spend here, the more time we spend building AutoGPT.
+        Please confirm that the issue you have is described well and precise in the title above ⬆️.
+        A good rule of thumb: What would you type if you were searching for the issue?
         
-        Please help us help you:
-        - Does it work on `stable` branch (https://github.com/Torantulino/Auto-GPT/tree/stable)?
-        - Does it work on current `master` (https://github.com/Torantulino/Auto-GPT/tree/master)?
-        - Search for existing issues, "add comment" is tidier than "new issue"
-        - Ask on our Discord (https://discord.gg/autogpt)
+        For example:
+        BAD - my auto-gpt keeps looping
+        GOOD - After performing execute_python_file, auto-gpt goes into a loop where it keeps trying to execute the file.
+        
+        ⚠️ SUPER-busy repo, please help the volunteer maintainers.
+        The less time we spend here, the more time we can spend building AutoGPT.
+        
+        Please help us help you by following these steps:
+        - Search for existing issues, adding a comment when you have the same or similar issue is tidier than "new issue" and 
+          newer issues will not be reviewed earlier, this is dependent on the current priorities set by our wonderful team
+        - Ask on our Discord if your issue is known when you are unsure (https://discord.gg/autogpt)
         - Provide relevant info:
-          - Provide commit-hash (`git rev-parse HEAD` gets it)
-          - If it's a pip/packages issue, provide pip version, python version
-          - If it's a crash, provide traceback.
+          - Provide commit-hash (`git rev-parse HEAD` gets it) if possible
+          - If it's a pip/packages issue, mention this in the title and provide pip version, python version
+          - If it's a crash, provide traceback and describe the error you got as precise as possible in the title.
+
   - type: dropdown
     attributes:
       label: Which Operating System are you using?
@@ -54,9 +62,15 @@ body:
         - Docker
         - Devcontainer / Codespace
         - Windows Subsystem for Linux (WSL)
-        - Other (Please specify in your problem)
+        - Other
     validations:
       required: true
+      nested_fields:
+        - type: text
+          attributes:
+            label: Specify the system
+            description: Please specify the system you are working on.
+
   - type: dropdown
     attributes:
       label: Which version of Auto-GPT are you using?
@@ -71,61 +85,80 @@ body:
         - Master (branch)
     validations:
       required: true
+
   - type: dropdown
     attributes:
-      label: GPT-3 or GPT-4?
+      label: Do you use OpenAI GPT-3 or GPT-4?
       description: >
         If you are using Auto-GPT with `--gpt3only`, your problems may be caused by
         the [limitations](https://github.com/Significant-Gravitas/Auto-GPT/issues?q=is%3Aissue+label%3A%22AI+model+limitation%22) of GPT-3.5.
       options:
         - GPT-3.5
         - GPT-4
+        - GPT-4(32k)
     validations:
       required: true
-  - type: textarea
+
+  - type: dropdown
     attributes:
-      label: Steps to reproduce 🕹
-      description: |
-        **⚠️ Issues that we can't reproduce will be closed.**
-  - type: textarea
-    attributes:
-      label: Current behavior 😯
-      description: Describe what happens instead of the expected behavior.
-  - type: textarea
-    attributes:
-      label: Expected behavior 🤔
-      description: Describe what should happen.
-  - type: textarea
-    attributes:
-      label: Your prompt 📝
+      label: Which area covers your issue best?
       description: >
-        If applicable please provide the prompt you are using. Your prompt is stored in your `ai_settings.yaml` file.
-      value: |
-        ```yaml
-        # Paste your prompt here
-        ```
+        Select the area related to the issue you are reporting.
+      options:
+        - Installation and setup
+        - Memory
+        - Performance
+        - Prompt
+        - Commands
+        - Plugins
+        - AI Model Limitations
+        - Challenges
+        - Documentation
+        - Logging
+        - Agents
+        - Other
+    validations:
+      required: true
+      autolabels: true
+      nested_fields:
+        - type: text
+          attributes:
+            label: Specify the area
+            description: Please specify the area you think is best related to the issue.
+
   - type: textarea
     attributes:
-      label: Your Logs 📒
-      description: |
-        Please include the log showing your error and the command that caused it, if applicable.
-        You can copy it from your terminal or from `logs/activity.log`.
-        This will help us understand your issue better!
-        
-        <details>
-        <summary><i>Example</i></summary>
-        ```log
-        INFO NEXT ACTION:  COMMAND = execute_shell  ARGUMENTS = {'command_line': 'some_command'}
-        INFO -=-=-=-=-=-=-= COMMAND AUTHORISED BY USER -=-=-=-=-=-=-=
-        Traceback (most recent call last):
-        File "/home/anaconda3/lib/python3.9/site-packages/openai/api_requestor.py", line 619, in _interpret_response
-        self._interpret_response_line(
-        File "/home/anaconda3/lib/python3.9/site-packages/openai/api_requestor.py", line 682, in _interpret_response_line
-        raise self.handle_error_response(
-        openai.error.InvalidRequestError: This model's maximum context length is 8191 tokens, however you requested 10982 tokens (10982 in your prompt; 0 for the completion). Please reduce your prompt; or completion length.
-        ```
-        </details>
+      label: Describe your issue.
+      description: Describe the problem you are experiencing. Try to describe only the issue and phrase it short but clear. ⚠️ Provide NO other data in this field  
+    validations:
+      required: true
+
+  #Following are optional file content uploads
+  - type: markdown
+    attributes:
       value: |
-        ```log
-        <insert your logs here>
-        ```
+        ⚠️The following is OPTIONAL, please keep in mind that the log files may contain personal information such as credentials.⚠️
+        
+        "The log files are located in the folder 'logs' inside the main auto-gpt folder."
+        
+  - type: input
+    attributes:
+      label: Upload Activity Log Content
+      description: |
+        Upload the activity log content, this can help us understand the issue better. 
+        To do this, go to the folder logs in your main auto-gpt folder, open activity.log and copy/paste the contents to this field. 
+        ⚠️ The activity log may contain personal data given to auto-gpt by you in prompt or input as well as 
+        any personal information that auto-gpt collected out of files during last run. Do not add the activity log if you are not comfortable with sharing it. ⚠️
+    validations:
+      required: false
+
+  - type: input
+    attributes:
+      label: Upload Error Log Content
+      description: |
+        Upload the error log content, this will help us understand the issue better. 
+        To do this, go to the folder logs in your main auto-gpt folder, open error.log and copy/paste the contents to this field. 
+        ⚠️ The error log may contain personal data given to auto-gpt by you in prompt or input as well as 
+        any personal information that auto-gpt collected out of files during last run. Do not add the activity log if you are not comfortable with sharing it. ⚠️
+    validations:
+      required: false

.github/ISSUE_TEMPLATE/2.feature.yml

@@ -1,13 +1,12 @@
 name: Feature request 🚀
-description: Suggest a new idea for Auto-GPT.
+description: Suggest a new idea for Auto-GPT!
 labels: ['status: needs triage']
 body:
   - type: markdown
     attributes:
       value: |
+        First, check out our [wiki page on Contributing](https://github.com/Significant-Gravitas/Nexus/wiki/Contributing)
         Please provide a searchable summary of the issue in the title above ⬆️.
-
-        Thanks for contributing by creating an issue! ❤️
   - type: checkboxes
     attributes:
       label: Duplicates
@@ -26,4 +25,4 @@ body:
   - type: textarea
     attributes:
       label: Motivation 🔦
-      description: What are you trying to accomplish? How has the lack of this feature affected you? Providing context helps us come up with a solution that is more useful in the real world.
+      description: What are you trying to accomplish? How has the lack of this feature affected you? Providing context helps us come up with a solution that is more useful in the real world.

.github/PULL_REQUEST_TEMPLATE.md

@@ -14,6 +14,8 @@ Provide clear documentation and explanations of the changes made.
 Ensure diffs are limited to the intended lines — no applying preferred formatting styles or line endings (unless that's what the PR is about).
 For guidance on committing only the specific lines you have changed, refer to this helpful video: https://youtu.be/8-hSNHHbiZg
 
+Check out our [wiki page on Contributing](https://github.com/Significant-Gravitas/Nexus/wiki/Contributing)
+
 By following these guidelines, your PRs are more likely to be merged quickly after testing, as long as they align with the project's overall direction. -->
 
 ### Background
@@ -33,7 +35,14 @@ By following these guidelines, your PRs are more likely to be merged quickly aft
 - [ ] I have thoroughly tested my changes with multiple different prompts.
 - [ ] I have considered potential risks and mitigations for my changes.
 - [ ] I have documented my changes clearly and comprehensively.
-- [ ] I have not snuck in any "extra" small tweaks changes <!-- Submit these as separate Pull Requests, they are the easiest to merge! -->
+- [ ] I have not snuck in any "extra" small tweaks changes. <!-- Submit these as separate Pull Requests, they are the easiest to merge! -->
+- [ ] I have run the following commands against my code to ensure it passes our linters:
+    ```shell
+    black .
+    isort .
+    mypy
+    autoflake --remove-all-unused-imports --recursive --ignore-init-module-imports --ignore-pass-after-docstring autogpt tests --in-place
+    ```
 
 <!-- If you haven't added tests, please explain why. If you have, check the appropriate box. If you've ensured your PR is atomic and well-documented, check the corresponding boxes. -->
 

.github/workflows/benchmarks.yml

@@ -1,31 +1,73 @@
-name: Run Benchmarks
+name: Benchmarks
 
 on:
+  schedule:
+    - cron: '0 8 * * *'
   workflow_dispatch:
 
 jobs:
-  build:
+  Benchmark:
+    name: ${{ matrix.config.task-name }}
     runs-on: ubuntu-latest
-
-    env:
-      python-version: '3.10'
+    timeout-minutes: 30
+    strategy:
+      fail-fast: false
+      matrix:
+        config:
+          - python-version: "3.10"
+            task: "tests/challenges"
+            task-name: "Mandatory Tasks"
+          - python-version: "3.10"
+            task: "--beat-challenges -ra tests/challenges"
+            task-name: "Challenging Tasks"
 
     steps:
-    - name: Checkout repository
-      uses: actions/checkout@v3
+      - name: Checkout repository
+        uses: actions/checkout@v3
+        with:
+          ref: master
 
-    - name: Set up Python ${{ env.python-version }}
-      uses: actions/setup-python@v4
-      with:
-        python-version: ${{ env.python-version }}
+      - name: Set up Python ${{ matrix.python-version }}
+        uses: actions/setup-python@v2
+        with:
+          python-version: ${{ matrix.config.python-version }}
 
-    - name: Install dependencies
-      run: |
-        python -m pip install --upgrade pip
-        pip install -r requirements.txt
+      - id: get_date
+        name: Get date
+        run: echo "date=$(date +'%Y-%m-%d')" >> $GITHUB_OUTPUT
 
-    - name: benchmark
-      env:
-        OPENAI_API_KEY: ${{ secrets.OPENAI_API_KEY }}
-      run: |
-         python benchmark/benchmark_entrepreneur_gpt_with_undecisive_user.py
+      - name: Set up Python dependency cache
+        uses: actions/cache@v3
+        with:
+          path: ~/.cache/pip
+          key: ${{ runner.os }}-pip-${{ hashFiles('**/requirements.txt') }}-${{ steps.get_date.outputs.date }}
+
+      - name: Install Python dependencies
+        run: |
+          python -m pip install --upgrade pip
+          pip install -r requirements.txt
+
+      - name: Run pytest with coverage
+        run: |
+          rm -rf tests/Auto-GPT-test-cassettes
+          pytest -n auto --record-mode=all ${{ matrix.config.task }}
+        env:
+          CI: true
+          PROXY: ${{ secrets.PROXY }}
+          AGENT_MODE: ${{ secrets.AGENT_MODE }}
+          AGENT_TYPE: ${{ secrets.AGENT_TYPE }}
+          PLAIN_OUTPUT: True
+
+      - name: Upload logs as artifact
+        if: always()
+        uses: actions/upload-artifact@v3
+        with:
+          name: test-logs-${{ matrix.config.task-name }}
+          path: logs/
+
+      - name: Upload cassettes as artifact
+        if: always()
+        uses: actions/upload-artifact@v3
+        with:
+          name: cassettes-${{ matrix.config.task-name }}
+          path: tests/Auto-GPT-test-cassettes/

.github/workflows/ci.yml

@@ -2,16 +2,24 @@ name: Python CI
 
 on:
   push:
-    branches: [ master ]
+    branches: [ master, ci-test* ]
+    paths-ignore:
+      - 'tests/Auto-GPT-test-cassettes'
+      - 'tests/challenges/current_score.json'
   pull_request:
-    branches: [ master ]
+    branches: [ stable, master, release-* ]
+  pull_request_target:
+    branches: [ master, release-*, ci-test* ]
 
 concurrency:
-  group: ${{ format('ci-{0}', github.head_ref && format('pr-{0}', github.event.pull_request.number) || github.sha) }}
-  cancel-in-progress: ${{ github.event_name == 'pull_request' }}
+  group: ${{ format('ci-{0}', github.head_ref && format('{0}-{1}', github.event_name, github.event.pull_request.number) || github.sha) }}
+  cancel-in-progress: ${{ startsWith(github.event_name, 'pull_request') }}
 
 jobs:
   lint:
+    # eliminate duplicate runs
+    if: github.event_name == 'push' || (github.event.pull_request.head.repo.fork == (github.event_name == 'pull_request_target'))
+
     runs-on: ubuntu-latest
     env:
       min-python-version: "3.10"
@@ -19,12 +27,26 @@ jobs:
     steps:
       - name: Checkout repository
         uses: actions/checkout@v3
+        with:
+          fetch-depth: 0
+          ref: ${{ github.event.pull_request.head.ref }}
+          repository: ${{ github.event.pull_request.head.repo.full_name }}
 
       - name: Set up Python ${{ env.min-python-version }}
         uses: actions/setup-python@v2
         with:
           python-version: ${{ env.min-python-version }}
 
+      - id: get_date
+        name: Get date
+        run: echo "date=$(date +'%Y-%m-%d')" >> $GITHUB_OUTPUT
+
+      - name: Set up Python dependency cache
+        uses: actions/cache@v3
+        with:
+          path: ~/.cache/pip
+          key: ${{ runner.os }}-pip-${{ hashFiles('**/requirements.txt') }}-${{ steps.get_date.outputs.date }}
+
       - name: Install dependencies
         run: |
           python -m pip install --upgrade pip
@@ -41,7 +63,19 @@ jobs:
         run: isort . --check
         if: success() || failure()
 
+      - name: Check mypy formatting
+        run: mypy
+        if: success() || failure()
+
+      - name: Check for unused imports and pass statements
+        run: |
+          cmd="autoflake --remove-all-unused-imports --recursive --ignore-init-module-imports --ignore-pass-after-docstring autogpt tests"
+          $cmd --check || (echo "You have unused imports or pass statements, please run '${cmd} --in-place'" && exit 1)
+
   test:
+    # eliminate duplicate runs
+    if: github.event_name == 'push' || (github.event.pull_request.head.repo.fork == (github.event_name == 'pull_request_target'))
+
     permissions:
       # Gives the action the necessary permissions for publishing new
       # comments in pull requests.
@@ -51,27 +85,170 @@ jobs:
       # comments (to avoid publishing multiple comments in the same PR)
       contents: write
     runs-on: ubuntu-latest
+    timeout-minutes: 30
     strategy:
       matrix:
-        python-version: ["3.10", "3.11"]
+        python-version: ["3.10"]
 
     steps:
-      - name: Check out repository
+      - name: Checkout repository
         uses: actions/checkout@v3
+        with:
+          fetch-depth: 0
+          ref: ${{ github.event.pull_request.head.ref }}
+          repository: ${{ github.event.pull_request.head.repo.full_name }}
+          submodules: true
+
+      - name: Configure git user Auto-GPT-Bot
+        run: |
+          git config --global user.name "Auto-GPT-Bot"
+          git config --global user.email "github-bot@agpt.co"
+
+      - name: Checkout cassettes
+        if: ${{ startsWith(github.event_name, 'pull_request') }}
+        run: |
+          cassette_branch="${{ github.event.pull_request.user.login }}-${{ github.event.pull_request.head.ref }}"
+          cd tests/Auto-GPT-test-cassettes
+
+          if git ls-remote --exit-code --heads origin $cassette_branch ; then
+            git fetch origin $cassette_branch
+            git fetch origin ${{ github.event.pull_request.base.ref }}
+
+            git checkout $cassette_branch
+
+            # Pick non-conflicting cassette updates from the base branch
+            git merge --no-commit --strategy-option=ours origin/${{ github.event.pull_request.base.ref }}
+            echo "Using cassettes from mirror branch '$cassette_branch'," \
+              "synced to upstream branch '${{ github.event.pull_request.base.ref }}'."
+          else
+            git checkout -b $cassette_branch
+            echo "Branch '$cassette_branch' does not exist in cassette submodule." \
+              "Using cassettes from '${{ github.event.pull_request.base.ref }}'."
+          fi
 
       - name: Set up Python ${{ matrix.python-version }}
         uses: actions/setup-python@v2
         with:
           python-version: ${{ matrix.python-version }}
 
-      - name: Install dependencies
+      - id: get_date
+        name: Get date
+        run: echo "date=$(date +'%Y-%m-%d')" >> $GITHUB_OUTPUT
+
+      - name: Set up Python dependency cache
+        uses: actions/cache@v3
+        with:
+          path: ~/.cache/pip
+          key: ${{ runner.os }}-pip-${{ hashFiles('**/requirements.txt') }}-${{ steps.get_date.outputs.date }}
+
+      - name: Install Python dependencies
         run: |
           python -m pip install --upgrade pip
           pip install -r requirements.txt
 
-      - name: Run unittest tests with coverage
+      - name: Run pytest with coverage
         run: |
-          pytest --cov=autogpt --cov-report term-missing --cov-branch --cov-report xml --cov-report term
+          pytest -n auto --cov=autogpt --cov-branch --cov-report term-missing --cov-report xml \
+            tests/unit tests/integration tests/challenges
+          python tests/challenges/utils/build_current_score.py
+        env:
+          CI: true
+          PROXY: ${{ secrets.PROXY }}
+          AGENT_MODE: ${{ secrets.AGENT_MODE }}
+          AGENT_TYPE: ${{ secrets.AGENT_TYPE }}
+          PLAIN_OUTPUT: True
 
       - name: Upload coverage reports to Codecov
         uses: codecov/codecov-action@v3
+
+      - id: setup_git_auth
+        name: Set up git token authentication
+        # Cassettes may be pushed even when tests fail
+        if: success() || failure()
+        run: |
+          config_key="http.${{ github.server_url }}/.extraheader"
+          base64_pat=$(echo -n "pat:${{ secrets.PAT_REVIEW }}" | base64 -w0)
+
+          git config "$config_key" \
+            "Authorization: Basic $base64_pat"
+
+          cd tests/Auto-GPT-test-cassettes
+          git config "$config_key" \
+            "Authorization: Basic $base64_pat"
+
+          echo "config_key=$config_key" >> $GITHUB_OUTPUT
+
+      - name: Push updated challenge scores
+        if: github.event_name == 'push'
+        run: |
+          score_file="tests/challenges/current_score.json"
+
+          if ! git diff --quiet $score_file; then
+            git add $score_file
+            git commit -m "Update challenge scores"
+            git push origin HEAD:${{ github.ref_name }}
+          else
+            echo "The challenge scores didn't change."
+          fi
+
+      - id: push_cassettes
+        name: Push updated cassettes
+        # For pull requests, push updated cassettes even when tests fail
+        if: github.event_name == 'push' || success() || failure()
+        run: |
+          if [ "${{ startsWith(github.event_name, 'pull_request') }}" = "true" ]; then
+            is_pull_request=true
+            cassette_branch="${{ github.event.pull_request.user.login }}-${{ github.event.pull_request.head.ref }}"
+          else
+            cassette_branch="${{ github.ref_name }}"
+          fi
+
+          cd tests/Auto-GPT-test-cassettes
+          # Commit & push changes to cassettes if any
+          if ! git diff --quiet; then
+            git add .
+            git commit -m "Auto-update cassettes"
+            git push origin HEAD:$cassette_branch
+            if [ ! $is_pull_request ]; then
+              cd ../..
+              git add tests/Auto-GPT-test-cassettes
+              git commit -m "Update cassette submodule"
+              git push origin HEAD:$cassette_branch
+            fi
+            echo "updated=true" >> $GITHUB_OUTPUT
+          else
+            echo "updated=false" >> $GITHUB_OUTPUT
+            echo "No cassette changes to commit"
+          fi
+
+      - name: Post Set up git token auth
+        if: steps.setup_git_auth.outcome == 'success'
+        run: |
+          git config --unset-all '${{ steps.setup_git_auth.outputs.config_key }}'
+          git submodule foreach git config --unset-all '${{ steps.setup_git_auth.outputs.config_key }}'
+
+      - name: Apply "behaviour change" label and comment on PR
+        if: ${{ startsWith(github.event_name, 'pull_request') }}
+        run: |
+          PR_NUMBER=${{ github.event.pull_request.number }}
+          TOKEN=${{ secrets.PAT_REVIEW }}
+          REPO=${{ github.repository }}
+
+          if [[ "${{ steps.push_cassettes.outputs.updated }}" == "true" ]]; then
+            echo "Adding label and comment..."
+            curl -X POST \
+            -H "Authorization: Bearer $TOKEN" \
+            -H "Accept: application/vnd.github.v3+json" \
+            https://api.github.com/repos/$REPO/issues/$PR_NUMBER/labels \
+            -d '{"labels":["behaviour change"]}'
+
+            echo $TOKEN | gh auth login --with-token
+            gh api repos/$REPO/issues/$PR_NUMBER/comments -X POST -F body="You changed AutoGPT's behaviour. The cassettes have been updated and will be merged to the submodule when this Pull Request gets merged."
+          fi
+
+      - name: Upload logs as artifact
+        if: always()
+        uses: actions/upload-artifact@v3
+        with:
+          name: test-logs
+          path: logs/

.github/workflows/docker-ci.yml

@@ -3,8 +3,11 @@ name: Docker CI
 on:
   push:
     branches: [ master ]
+    paths-ignore:
+      - 'tests/Auto-GPT-test-cassettes'
+      - 'tests/challenges/current_score.json'
   pull_request:
-    branches: [ master ]
+    branches: [ master, release-*, stable ]
 
 concurrency:
   group: ${{ format('docker-ci-{0}', github.head_ref && format('pr-{0}', github.event.pull_request.number) || github.sha) }}
@@ -73,43 +76,51 @@ jobs:
   # Docker setup needs fixing before this is going to work: #1843
   test:
     runs-on: ubuntu-latest
+    timeout-minutes: 30
     needs: build
     steps:
-    - name: Checkout repository
-      uses: actions/checkout@v3
+      - name: Check out repository
+        uses: actions/checkout@v3
+        with:
+          fetch-depth: 0
+          submodules: true
 
-    - name: Set up Docker Buildx
-      uses: docker/setup-buildx-action@v2
+      - name: Set up Docker Buildx
+        uses: docker/setup-buildx-action@v2
 
-    - id: build
-      name: Build image
-      uses: docker/build-push-action@v3
-      with:
-        build-args: BUILD_TYPE=dev  # include pytest
-        tags: ${{ env.IMAGE_NAME }}
-        load: true                  # save to docker images
-        # cache layers in GitHub Actions cache to speed up builds
-        cache-from: type=gha,scope=docker-dev
-        cache-to: type=gha,scope=docker-dev,mode=max
+      - id: build
+        name: Build image
+        uses: docker/build-push-action@v3
+        with:
+          build-args: BUILD_TYPE=dev  # include pytest
+          tags: ${{ env.IMAGE_NAME }}
+          load: true                  # save to docker images
+          # cache layers in GitHub Actions cache to speed up builds
+          cache-from: type=gha,scope=docker-dev
+          cache-to: type=gha,scope=docker-dev,mode=max
 
-    - id: test
-      name: Run tests
-      env:
-        CI: true
-        OPENAI_API_KEY: ${{ secrets.OPENAI_API_KEY }}
-      run: |
-        set +e
-        test_output=$(
-          docker run --env CI --env OPENAI_API_KEY --entrypoint python ${{ env.IMAGE_NAME }} -m \
-          pytest --cov=autogpt --cov-report term-missing --cov-branch --cov-report xml --cov-report term 2>&1
-        )
-        test_failure=$?
+      - id: test
+        name: Run tests
+        env:
+          PLAIN_OUTPUT: True
+          CI: true
+          OPENAI_API_KEY: ${{ secrets.OPENAI_API_KEY }}
+        run: |
+          set +e
+          test_output=$(
+            docker run --env CI --env OPENAI_API_KEY --entrypoint python ${{ env.IMAGE_NAME }} -m \
+            pytest -n auto --cov=autogpt --cov-branch --cov-report term-missing \
+              tests/unit tests/integration 2>&1
+          )
+          test_failure=$?
 
-        echo "$test_output"
+          echo "$test_output"
 
-        cat << $EOF >> $GITHUB_STEP_SUMMARY
-        # Tests $([ $test_failure = 0 ] && echo '✅' || echo '❌')
-        \`\`\`
-        $test_output
-        \`\`\`
-        $EOF
+          cat << $EOF >> $GITHUB_STEP_SUMMARY
+          # Tests $([ $test_failure = 0 ] && echo '✅' || echo '❌')
+          \`\`\`
+          $test_output
+          \`\`\`
+          $EOF
+
+          exit $test_failure

.github/workflows/pr-label.yml

@@ -3,7 +3,10 @@ name: "Pull Request auto-label"
 on:
   # So that PRs touching the same files as the push are updated
   push:
-    branches: [ master ]
+    branches: [ master, release-* ]
+    paths-ignore:
+      - 'tests/Auto-GPT-test-cassettes'
+      - 'tests/challenges/current_score.json'
   # So that the `dirtyLabel` is removed if conflicts are resolve
   # We recommend `pull_request_target` so that github secrets are available.
   # In `pull_request` we wouldn't be able to change labels of fork PRs
@@ -45,11 +48,10 @@ jobs:
           s_label: 'size/s'
           s_max_size: 10
           m_label: 'size/m'
-          m_max_size: 50
+          m_max_size: 100
           l_label: 'size/l'
-          l_max_size: 200
+          l_max_size: 500
           xl_label: 'size/xl'
           message_if_xl: >
-            This PR exceeds the recommended size of 200 lines.
+            This PR exceeds the recommended size of 500 lines.
             Please make sure you are NOT addressing multiple issues with one PR.
-            Note this PR might be rejected due to its size  

.github/workflows/sponsors_readme.yml

@@ -1,28 +0,0 @@
-name: Generate Sponsors README
-
-on:
-  workflow_dispatch:
-  schedule:
-    - cron: '0 */12 * * *'
-
-jobs:
-  deploy:
-    runs-on: ubuntu-latest
-    steps:
-      - name: Checkout 🛎️
-        uses: actions/checkout@v3
-
-      - name: Generate Sponsors 💖
-        uses: JamesIves/github-sponsors-readme-action@v1
-        with:
-          token: ${{ secrets.README_UPDATER_PAT }}
-          file: 'README.md'
-          minimum: 2500
-          maximum: 99999
-
-      - name: Deploy to GitHub Pages 🚀
-        uses: JamesIves/github-pages-deploy-action@v4
-        with:
-          branch: master
-          folder: '.'
-          token: ${{ secrets.README_UPDATER_PAT }}

.gitignore

@@ -1,12 +1,7 @@
 ## Original ignores
 autogpt/keys.py
-autogpt/*json
-autogpt/node_modules/
-autogpt/__pycache__/keys.cpython-310.pyc
-autogpt/auto_gpt_workspace
-package-lock.json
-*.pyc
-auto_gpt_workspace/*
+autogpt/*.json
+**/auto_gpt_workspace/*
 *.mpeg
 .env
 azure.yaml
@@ -21,6 +16,7 @@ logs
 *.log
 *.mp3
 mem.sqlite3
+venvAutoGPT
 
 # Byte-compiled / optimized / DLL files
 __pycache__/
@@ -36,6 +32,7 @@ build/
 develop-eggs/
 dist/
 plugins/
+plugins_config.yaml
 downloads/
 eggs/
 .eggs/
@@ -162,4 +159,4 @@ vicuna-*
 openai/
 
 # news
-CURRENT_BULLETIN.md
+CURRENT_BULLETIN.md

.gitmodules

@@ -0,0 +1,4 @@
+[submodule "tests/Auto-GPT-test-cassettes"]
+    path = tests/Auto-GPT-test-cassettes
+    url = https://github.com/Significant-Gravitas/Auto-GPT-test-cassettes
+    branch = master

.pre-commit-config.yaml

@@ -22,8 +22,18 @@ repos:
       - id: black
         language_version: python3.10
 
+  - repo: https://github.com/pre-commit/mirrors-mypy
+    rev: 'v1.3.0'
+    hooks:
+      - id: mypy
+
   - repo: local
     hooks:
+      - id: autoflake
+        name: autoflake
+        entry: autoflake --in-place --remove-all-unused-imports --recursive --ignore-init-module-imports --ignore-pass-after-docstring autogpt tests
+        language: python
+        types: [ python ]
       - id: pytest-check
         name: pytest-check
         entry: pytest --cov=autogpt --without-integration --without-slow-integration

BULLETIN.md

@@ -1,9 +1,27 @@
-Welcome to Auto-GPT! We'll keep you informed of the latest news and features by printing messages here.
-If you don't wish to see this message, you can run Auto-GPT with the --skip-news flag
+# Website and Documentation Site 📰📖
+Check out *https://agpt.co*, the official news & updates site for Auto-GPT!
+The documentation also has a place here, at *https://docs.agpt.co*
 
-# INCLUDED COMMAND 'send_tweet' IS DEPRICATED, AND WILL BE REMOVED IN THE NEXT STABLE RELEASE
-Base Twitter functionality (and more) is now covered by plugins: https://github.com/Significant-Gravitas/Auto-GPT-Plugins
+# For contributors 👷🏼
+Since releasing v0.3.0, whave been working on re-architecting the Auto-GPT core to make it more extensible and make room for structural performance-oriented R&D.
 
-## Changes to Docker configuration
-The workdir has been changed from /home/appuser to /app. Be sure to update any volume mounts accordingly.
+Check out the contribution guide on our wiki:
+https://github.com/Significant-Gravitas/Auto-GPT/wiki/Contributing
 
+# 🚀 v0.4.1 Release 🚀
+Two weeks and 50+ pull requests have passed since v0.4.0, and we are happy to announce the release of v0.4.1!
+
+Highlights and notable changes since v0.4.0:
+- The .env.template is more readable and better explains the purpose of each environment variable.
+- More dependable search
+    - The CUSTOM_SEARCH_ENGINE_ID variable has been replaced to GOOGLE_CUSTOM_SEARCH_ENGINE_ID, make sure you update it.
+- Better read_file
+- More reliable python code execution
+- Lots of JSON error fixes
+- Directory-based plugins
+
+## Further fixes and changes 🛠️
+Under the hood, we've done a bunch of work improving architectures and streamlining code. Most of that won't be user-visible
+
+## Take a look at the Release Notes on Github for the full changelog!
+https://github.com/Significant-Gravitas/Auto-GPT/releases

CONTRIBUTING.md

@@ -1,148 +1,14 @@
-# Contributing to Auto-GPT
+We maintain a knowledgebase at this [wiki](https://github.com/Significant-Gravitas/Nexus/wiki)
 
-First of all, thank you for considering contributing to our project! We appreciate your time and effort, and we value any contribution, whether it's reporting a bug, suggesting a new feature, or submitting a pull request.
+We would like to say "We value all contributions". After all, we are an open-source project, so we should say something fluffy like this, right?
 
-This document provides guidelines and best practices to help you contribute effectively.
+However the reality is that some contributions are SUPER-valuable, while others create more trouble than they are worth and actually _create_ work for the core team.
 
-## Code of Conduct
+If you wish to contribute, please look through the wiki [contributing](https://github.com/Significant-Gravitas/Nexus/wiki/Contributing) page.
 
-By participating in this project, you agree to abide by our [Code of Conduct]. Please read it to understand the expectations we have for everyone who contributes to this project.
+If you wish to involve with the project (beyond just contributing PRs), please read the wiki [catalyzing](https://github.com/Significant-Gravitas/Nexus/wiki/Catalyzing) page.
 
-[Code of Conduct]: https://significant-gravitas.github.io/Auto-GPT/code-of-conduct.md
+In fact, why not just look through the whole wiki (it's only a few pages) and hop on our discord (you'll find it in the wiki).
 
-## 📢 A Quick Word
-Right now we will not be accepting any Contributions that add non-essential commands to Auto-GPT.
-
-However, you absolutely can still add these commands to Auto-GPT in the form of plugins.
-Please check out this [template](https://github.com/Significant-Gravitas/Auto-GPT-Plugin-Template).
-
-## Getting Started
-
-1. Fork the repository and clone your fork.
-2. Create a new branch for your changes (use a descriptive name, such as `fix-bug-123` or `add-new-feature`).
-3. Make your changes in the new branch.
-4. Test your changes thoroughly.
-5. Commit and push your changes to your fork.
-6. Create a pull request following the guidelines in the [Submitting Pull Requests](#submitting-pull-requests) section.
-
-## How to Contribute
-
-### Reporting Bugs
-
-If you find a bug in the project, please create an issue on GitHub with the following information:
-
-- A clear, descriptive title for the issue.
-- A description of the problem, including steps to reproduce the issue.
-- Any relevant logs, screenshots, or other supporting information.
-
-### Suggesting Enhancements
-
-If you have an idea for a new feature or improvement, please create an issue on GitHub with the following information:
-
-- A clear, descriptive title for the issue.
-- A detailed description of the proposed enhancement, including any benefits and potential drawbacks.
-- Any relevant examples, mockups, or supporting information.
-
-### Submitting Pull Requests
-
-When submitting a pull request, please ensure that your changes meet the following criteria:
-
-- Your pull request should be atomic and focus on a single change.
-- Your pull request should include tests for your change. We automatically enforce this with [CodeCov](https://docs.codecov.com/docs/commit-status)
-- You should have thoroughly tested your changes with multiple different prompts.
-- You should have considered potential risks and mitigations for your changes.
-- You should have documented your changes clearly and comprehensively.
-- You should not include any unrelated or "extra" small tweaks or changes.
-
-## Style Guidelines
-
-### Code Formatting
-
-We use the `black` and `isort` code formatters to maintain a consistent coding style across the project. Please ensure that your code is formatted properly before submitting a pull request.
-
-To format your code, run the following commands in the project's root directory:
-
-```bash
-python -m black .
-python -m isort .
-```
-
-Or if you have these tools installed globally:
-```bash
-black .
-isort .
-```
-
-### Pre-Commit Hooks
-
-We use pre-commit hooks to ensure that code formatting and other checks are performed automatically before each commit. To set up pre-commit hooks for this project, follow these steps:
-
-Install the pre-commit package using pip:
-```bash
-pip install pre-commit
-```
-
-Run the following command in the project's root directory to install the pre-commit hooks:
-```bash
-pre-commit install
-```
-
-Now, the pre-commit hooks will run automatically before each commit, checking your code formatting and other requirements.
-
-If you encounter any issues or have questions, feel free to reach out to the maintainers or open a new issue on GitHub. We're here to help and appreciate your efforts to contribute to the project.
-
-Happy coding, and once again, thank you for your contributions!
-
-Maintainers will look at PR that have no merge conflicts when deciding what to add to the project. Make sure your PR shows up here:
-https://github.com/Significant-Gravitas/Auto-GPT/pulls?q=is%3Apr+is%3Aopen+-label%3Aconflicts
-
-## Testing your changes
-
-If you add or change code, make sure the updated code is covered by tests.
-To increase coverage if necessary, [write tests using pytest].
-
-For more info on running tests, please refer to ["Running tests"](https://significant-gravitas.github.io/Auto-GPT/testing/).
-
-[write tests using pytest]: https://realpython.com/pytest-python-testing/
-
-### API-dependent tests
-
-To run tests that involve making calls to the OpenAI API, we use VCRpy. It caches known
-requests and matching responses in so-called *cassettes*, allowing us to run the tests
-in CI without needing actual API access.
-
-When changes cause a test prompt to be generated differently, it will likely miss the
-cache and make a request to the API, updating the cassette with the new request+response.
-*Be sure to include the updated cassette in your PR!*
-
-When you run Pytest locally:
-
-- If no prompt change: you will not consume API tokens because there are no new OpenAI calls required.
-- If the prompt changes in a way that the cassettes are not reusable:
-    - If no API key, the test fails. It requires a new cassette. So, add an API key to .env.
-    - If the API key is present, the tests will make a real call to OpenAI.
-        - If the test ends up being successful, your prompt changes didn't introduce regressions. This is good. Commit your cassettes to your PR.
-        - If the test is unsuccessful:
-            - Either: Your change made Auto-GPT less capable, in that case, you have to change your code.
-            - Or: The test might be poorly written. In that case, you can make suggestions to change the test.
-
-In our CI pipeline, Pytest will use the cassettes and not call paid API providers, so we need your help to record the replays that you break.
-
-
-### Community Challenges
-Challenges are goals we need Auto-GPT to achieve. 
-To pick the challenge you like, go to the tests/integration/challenges folder and select the areas you would like to work on. 
-- a challenge is new if level_currently_beaten is None
-- a challenge is in progress if level_currently_beaten is greater or equal to 1
-- a challenge is beaten if level_currently_beaten = max_level
-
-Here is an example of how to run the memory challenge A and attempt to beat level 3.
-
-pytest -s tests/integration/challenges/memory/test_memory_challenge_a.py --level=3
-
-To beat a challenge, you're not allowed to change anything in the tests folder, you have to add code in the autogpt folder
-
-Challenges use cassettes. Cassettes allow us to replay your runs in our CI pipeline.
-Don't hesitate to delete the cassettes associated to the challenge you're working on if you need to. Otherwise it will keep replaying the last run.
-
-Once you've beaten a new level of a challenge, please create a pull request and we will analyze how you changed Auto-GPT to beat the challenge.
+❤️ & 🔆
+The team @ Auto-GPT

Dockerfile

@@ -6,11 +6,13 @@ FROM python:3.10-slim AS autogpt-base
 
 # Install browsers
 RUN apt-get update && apt-get install -y \
-    chromium-driver firefox-esr \
-    ca-certificates
+    chromium-driver firefox-esr ca-certificates \
+    && apt-get clean && rm -rf /var/lib/apt/lists/*
 
 # Install utilities
-RUN apt-get install -y curl jq wget git
+RUN apt-get update && apt-get install -y \
+    curl jq wget git \
+    && apt-get clean && rm -rf /var/lib/apt/lists/*
 
 # Set environment variables
 ENV PIP_NO_CACHE_DIR=yes \
@@ -22,7 +24,7 @@ ENV PATH="$PATH:/root/.local/bin"
 COPY requirements.txt .
 
 # Set the entrypoint
-ENTRYPOINT ["python", "-m", "autogpt"]
+ENTRYPOINT ["python", "-m", "autogpt", "--install-plugin-deps"]
 
 # dev build -> include everything
 FROM autogpt-base as autogpt-dev
@@ -36,5 +38,9 @@ RUN sed -i '/Items below this point will not be included in the Docker Image/,$d
 	pip install --no-cache-dir -r requirements.txt
 WORKDIR /app
 ONBUILD COPY autogpt/ ./autogpt
+ONBUILD COPY scripts/ ./scripts
+ONBUILD COPY plugins/ ./plugins
+ONBUILD COPY prompt_settings.yaml ./prompt_settings.yaml
+ONBUILD RUN mkdir ./data
 
 FROM autogpt-${BUILD_TYPE} AS auto-gpt

autogpt/agent/agent.py

@@ -1,11 +1,29 @@
+import json
+import signal
+import sys
+from datetime import datetime
+
 from colorama import Fore, Style
 
-from autogpt.app import execute_command, get_command
+from autogpt.commands.command import CommandRegistry
 from autogpt.config import Config
-from autogpt.json_utils.json_fix_llm import fix_json_using_multiple_techniques
-from autogpt.json_utils.utilities import LLM_DEFAULT_RESPONSE_FORMAT, validate_json
-from autogpt.llm import chat_with_ai, create_chat_completion, create_chat_message
+from autogpt.config.ai_config import AIConfig
+from autogpt.json_utils.utilities import extract_json_from_response, validate_json
+from autogpt.llm.base import ChatSequence
+from autogpt.llm.chat import chat_with_ai, create_chat_completion
+from autogpt.llm.providers.openai import OPEN_AI_CHAT_MODELS
+from autogpt.llm.utils import count_string_tokens
+from autogpt.log_cycle.log_cycle import (
+    FULL_MESSAGE_HISTORY_FILE_NAME,
+    NEXT_ACTION_FILE_NAME,
+    PROMPT_SUPERVISOR_FEEDBACK_FILE_NAME,
+    SUPERVISOR_FEEDBACK_FILE_NAME,
+    USER_INPUT_FILE_NAME,
+    LogCycleHandler,
+)
 from autogpt.logs import logger, print_assistant_thoughts
+from autogpt.memory.message_history import MessageHistory
+from autogpt.memory.vector import VectorMemory
 from autogpt.speech import say_text
 from autogpt.spinner import Spinner
 from autogpt.utils import clean_input
@@ -18,7 +36,6 @@ class Agent:
     Attributes:
         ai_name: The name of the agent.
         memory: The memory object to use.
-        full_message_history: The full message history.
         next_action_count: The number of actions to execute.
         system_prompt: The system prompt is the initial prompt that defines everything
           the AI needs to know to achieve its task successfully.
@@ -27,7 +44,7 @@ class Agent:
 
         triggering_prompt: The last sentence the AI will see before answering.
             For Auto-GPT, this prompt is:
-            Determine which next command to use, and respond using the format specified
+            Determine exactly one command to use, and respond using the format specified
               above:
             The triggering prompt is not part of the system prompt because between the
               system prompt and the triggering
@@ -43,110 +60,152 @@ class Agent:
 
     def __init__(
         self,
-        ai_name,
-        memory,
-        full_message_history,
-        next_action_count,
-        command_registry,
-        config,
-        system_prompt,
-        triggering_prompt,
-        workspace_directory,
+        ai_name: str,
+        memory: VectorMemory,
+        next_action_count: int,
+        command_registry: CommandRegistry,
+        ai_config: AIConfig,
+        system_prompt: str,
+        triggering_prompt: str,
+        workspace_directory: str,
+        config: Config,
     ):
-        cfg = Config()
         self.ai_name = ai_name
         self.memory = memory
-        self.summary_memory = (
-            "I was created."  # Initial memory necessary to avoid hilucination
-        )
-        self.last_memory_index = 0
-        self.full_message_history = full_message_history
+        self.history = MessageHistory(self)
         self.next_action_count = next_action_count
         self.command_registry = command_registry
         self.config = config
+        self.ai_config = ai_config
         self.system_prompt = system_prompt
         self.triggering_prompt = triggering_prompt
-        self.workspace = Workspace(workspace_directory, cfg.restrict_to_workspace)
+        self.workspace = Workspace(workspace_directory, config.restrict_to_workspace)
+        self.created_at = datetime.now().strftime("%Y%m%d_%H%M%S")
+        self.cycle_count = 0
+        self.log_cycle_handler = LogCycleHandler()
+        self.fast_token_limit = OPEN_AI_CHAT_MODELS.get(
+            config.fast_llm_model
+        ).max_tokens
 
     def start_interaction_loop(self):
+        # Avoid circular imports
+        from autogpt.app import execute_command, get_command
+
         # Interaction Loop
-        cfg = Config()
-        loop_count = 0
+        self.cycle_count = 0
         command_name = None
         arguments = None
         user_input = ""
 
+        # Signal handler for interrupting y -N
+        def signal_handler(signum, frame):
+            if self.next_action_count == 0:
+                sys.exit()
+            else:
+                print(
+                    Fore.RED
+                    + "Interrupt signal received. Stopping continuous command execution."
+                    + Style.RESET_ALL
+                )
+                self.next_action_count = 0
+
+        signal.signal(signal.SIGINT, signal_handler)
+
         while True:
             # Discontinue if continuous limit is reached
-            loop_count += 1
+            self.cycle_count += 1
+            self.log_cycle_handler.log_count_within_cycle = 0
+            self.log_cycle_handler.log_cycle(
+                self.ai_config.ai_name,
+                self.created_at,
+                self.cycle_count,
+                [m.raw() for m in self.history],
+                FULL_MESSAGE_HISTORY_FILE_NAME,
+            )
             if (
-                cfg.continuous_mode
-                and cfg.continuous_limit > 0
-                and loop_count > cfg.continuous_limit
+                self.config.continuous_mode
+                and self.config.continuous_limit > 0
+                and self.cycle_count > self.config.continuous_limit
             ):
                 logger.typewriter_log(
-                    "Continuous Limit Reached: ", Fore.YELLOW, f"{cfg.continuous_limit}"
+                    "Continuous Limit Reached: ",
+                    Fore.YELLOW,
+                    f"{self.config.continuous_limit}",
                 )
                 break
             # Send message to AI, get response
-            with Spinner("Thinking... "):
+            with Spinner("Thinking... ", plain_output=self.config.plain_output):
                 assistant_reply = chat_with_ai(
+                    self.config,
                     self,
                     self.system_prompt,
                     self.triggering_prompt,
-                    self.full_message_history,
-                    self.memory,
-                    cfg.fast_token_limit,
-                )  # TODO: This hardcodes the model to use GPT3.5. Make this an argument
+                    self.fast_token_limit,
+                    self.config.fast_llm_model,
+                )
 
-            assistant_reply_json = fix_json_using_multiple_techniques(assistant_reply)
-            for plugin in cfg.plugins:
+            try:
+                assistant_reply_json = extract_json_from_response(assistant_reply)
+                validate_json(assistant_reply_json)
+            except json.JSONDecodeError as e:
+                logger.error(f"Exception while validating assistant reply JSON: {e}")
+                assistant_reply_json = {}
+
+            for plugin in self.config.plugins:
                 if not plugin.can_handle_post_planning():
                     continue
-                assistant_reply_json = plugin.post_planning(self, assistant_reply_json)
+                assistant_reply_json = plugin.post_planning(assistant_reply_json)
 
             # Print Assistant thoughts
             if assistant_reply_json != {}:
-                validate_json(assistant_reply_json, LLM_DEFAULT_RESPONSE_FORMAT)
                 # Get command name and arguments
                 try:
                     print_assistant_thoughts(
-                        self.ai_name, assistant_reply_json, cfg.speak_mode
+                        self.ai_name, assistant_reply_json, self.config.speak_mode
                     )
                     command_name, arguments = get_command(assistant_reply_json)
-                    if cfg.speak_mode:
+                    if self.config.speak_mode:
                         say_text(f"I want to execute {command_name}")
 
                     arguments = self._resolve_pathlike_command_args(arguments)
 
                 except Exception as e:
                     logger.error("Error: \n", str(e))
+            self.log_cycle_handler.log_cycle(
+                self.ai_config.ai_name,
+                self.created_at,
+                self.cycle_count,
+                assistant_reply_json,
+                NEXT_ACTION_FILE_NAME,
+            )
 
-            if not cfg.continuous_mode and self.next_action_count == 0:
+            # First log new-line so user can differentiate sections better in console
+            logger.typewriter_log("\n")
+            logger.typewriter_log(
+                "NEXT ACTION: ",
+                Fore.CYAN,
+                f"COMMAND = {Fore.CYAN}{command_name}{Style.RESET_ALL}  "
+                f"ARGUMENTS = {Fore.CYAN}{arguments}{Style.RESET_ALL}",
+            )
+
+            if not self.config.continuous_mode and self.next_action_count == 0:
                 # ### GET USER AUTHORIZATION TO EXECUTE COMMAND ###
                 # Get key press: Prompt the user to press enter to continue or escape
                 # to exit
                 self.user_input = ""
-                logger.typewriter_log(
-                    "NEXT ACTION: ",
-                    Fore.CYAN,
-                    f"COMMAND = {Fore.CYAN}{command_name}{Style.RESET_ALL}  "
-                    f"ARGUMENTS = {Fore.CYAN}{arguments}{Style.RESET_ALL}",
-                )
-
                 logger.info(
-                    "Enter 'y' to authorise command, 'y -N' to run N continuous commands, 's' to run self-feedback commands"
+                    "Enter 'y' to authorise command, 'y -N' to run N continuous commands, 's' to run self-feedback commands, "
                     "'n' to exit program, or enter feedback for "
                     f"{self.ai_name}..."
                 )
                 while True:
-                    if cfg.chat_messages_enabled:
+                    if self.config.chat_messages_enabled:
                         console_input = clean_input("Waiting for your response...")
                     else:
                         console_input = clean_input(
                             Fore.MAGENTA + "Input:" + Style.RESET_ALL
                         )
-                    if console_input.lower().strip() == cfg.authorise_key:
+                    if console_input.lower().strip() == self.config.authorise_key:
                         user_input = "GENERATE NEXT COMMAND JSON"
                         break
                     elif console_input.lower().strip() == "s":
@@ -157,22 +216,22 @@ class Agent:
                         )
                         thoughts = assistant_reply_json.get("thoughts", {})
                         self_feedback_resp = self.get_self_feedback(
-                            thoughts, cfg.fast_llm_model
+                            thoughts, self.config.fast_llm_model
                         )
                         logger.typewriter_log(
                             f"SELF FEEDBACK: {self_feedback_resp}",
                             Fore.YELLOW,
                             "",
                         )
-                        if self_feedback_resp[0].lower().strip() == cfg.authorise_key:
-                            user_input = "GENERATE NEXT COMMAND JSON"
-                        else:
-                            user_input = self_feedback_resp
+                        user_input = self_feedback_resp
+                        command_name = "self_feedback"
                         break
                     elif console_input.lower().strip() == "":
                         logger.warn("Invalid input format.")
                         continue
-                    elif console_input.lower().startswith(f"{cfg.authorise_key} -"):
+                    elif console_input.lower().startswith(
+                        f"{self.config.authorise_key} -"
+                    ):
                         try:
                             self.next_action_count = abs(
                                 int(console_input.split(" ")[1])
@@ -185,12 +244,19 @@ class Agent:
                             )
                             continue
                         break
-                    elif console_input.lower() == cfg.exit_key:
+                    elif console_input.lower() == self.config.exit_key:
                         user_input = "EXIT"
                         break
                     else:
                         user_input = console_input
                         command_name = "human_feedback"
+                        self.log_cycle_handler.log_cycle(
+                            self.ai_config.ai_name,
+                            self.created_at,
+                            self.cycle_count,
+                            user_input,
+                            USER_INPUT_FILE_NAME,
+                        )
                         break
 
                 if user_input == "GENERATE NEXT COMMAND JSON":
@@ -203,37 +269,45 @@ class Agent:
                     logger.info("Exiting...")
                     break
             else:
-                # Print command
+                # First log new-line so user can differentiate sections better in console
+                logger.typewriter_log("\n")
+                # Print authorized commands left value
                 logger.typewriter_log(
-                    "NEXT ACTION: ",
-                    Fore.CYAN,
-                    f"COMMAND = {Fore.CYAN}{command_name}{Style.RESET_ALL}"
-                    f"  ARGUMENTS = {Fore.CYAN}{arguments}{Style.RESET_ALL}",
+                    f"{Fore.CYAN}AUTHORISED COMMANDS LEFT: {Style.RESET_ALL}{self.next_action_count}"
                 )
 
             # Execute command
             if command_name is not None and command_name.lower().startswith("error"):
-                result = (
-                    f"Command {command_name} threw the following error: {arguments}"
-                )
+                result = f"Could not execute command: {arguments}"
             elif command_name == "human_feedback":
                 result = f"Human feedback: {user_input}"
+            elif command_name == "self_feedback":
+                result = f"Self feedback: {user_input}"
             else:
-                for plugin in cfg.plugins:
+                for plugin in self.config.plugins:
                     if not plugin.can_handle_pre_command():
                         continue
                     command_name, arguments = plugin.pre_command(
                         command_name, arguments
                     )
                 command_result = execute_command(
-                    self.command_registry,
-                    command_name,
-                    arguments,
-                    self.config.prompt_generator,
+                    command_name=command_name,
+                    arguments=arguments,
+                    agent=self,
                 )
                 result = f"Command {command_name} returned: " f"{command_result}"
 
-                for plugin in cfg.plugins:
+                result_tlength = count_string_tokens(
+                    str(command_result), self.config.fast_llm_model
+                )
+                memory_tlength = count_string_tokens(
+                    str(self.history.summary_message()), self.config.fast_llm_model
+                )
+                if result_tlength + memory_tlength + 600 > self.fast_token_limit:
+                    result = f"Failure: command {command_name} returned too much output. \
+                        Do not execute this command again with the same arguments."
+
+                for plugin in self.config.plugins:
                     if not plugin.can_handle_post_command():
                         continue
                     result = plugin.post_command(command_name, result)
@@ -243,12 +317,10 @@ class Agent:
             # Check if there's a result from the command append it to the message
             # history
             if result is not None:
-                self.full_message_history.append(create_chat_message("system", result))
+                self.history.add("system", result, "action_result")
                 logger.typewriter_log("SYSTEM: ", Fore.YELLOW, result)
             else:
-                self.full_message_history.append(
-                    create_chat_message("system", "Unable to execute command")
-                )
+                self.history.add("system", "Unable to execute command", "action_result")
                 logger.typewriter_log(
                     "SYSTEM: ", Fore.YELLOW, "Unable to execute command"
                 )
@@ -276,15 +348,32 @@ class Agent:
         Returns:
             str: A feedback response generated using the provided thoughts dictionary.
         """
-        ai_role = self.config.ai_role
+        ai_role = self.ai_config.ai_role
 
-        feedback_prompt = f"Below is a message from an AI agent with the role of {ai_role}. Please review the provided Thought, Reasoning, Plan, and Criticism. If these elements accurately contribute to the successful execution of the assumed role, respond with the letter 'Y' followed by a space, and then explain why it is effective. If the provided information is not suitable for achieving the role's objectives, please provide one or more sentences addressing the issue and suggesting a resolution."
+        feedback_prompt = f"Below is a message from me, an AI Agent, assuming the role of {ai_role}. whilst keeping knowledge of my slight limitations as an AI Agent Please evaluate my thought process, reasoning, and plan, and provide a concise paragraph outlining potential improvements. Consider adding or removing ideas that do not align with my role and explaining why, prioritizing thoughts based on their significance, or simply refining my overall thought process."
         reasoning = thoughts.get("reasoning", "")
         plan = thoughts.get("plan", "")
         thought = thoughts.get("thoughts", "")
-        criticism = thoughts.get("criticism", "")
-        feedback_thoughts = thought + reasoning + plan + criticism
-        return create_chat_completion(
-            [{"role": "user", "content": feedback_prompt + feedback_thoughts}],
-            llm_model,
+        feedback_thoughts = thought + reasoning + plan
+
+        prompt = ChatSequence.for_model(llm_model)
+        prompt.add("user", feedback_prompt + feedback_thoughts)
+
+        self.log_cycle_handler.log_cycle(
+            self.ai_config.ai_name,
+            self.created_at,
+            self.cycle_count,
+            prompt.raw(),
+            PROMPT_SUPERVISOR_FEEDBACK_FILE_NAME,
         )
+
+        feedback = create_chat_completion(prompt)
+
+        self.log_cycle_handler.log_cycle(
+            self.ai_config.ai_name,
+            self.created_at,
+            self.cycle_count,
+            feedback,
+            SUPERVISOR_FEEDBACK_FILE_NAME,
+        )
+        return feedback

autogpt/agent/agent_manager.py

@@ -1,10 +1,9 @@
 """Agent manager for managing GPT agents"""
 from __future__ import annotations
 
-from typing import List
-
-from autogpt.config.config import Config
-from autogpt.llm import Message, create_chat_completion
+from autogpt.config import Config
+from autogpt.llm.base import ChatSequence
+from autogpt.llm.chat import Message, create_chat_completion
 from autogpt.singleton import Singleton
 
 
@@ -13,55 +12,55 @@ class AgentManager(metaclass=Singleton):
 
     def __init__(self):
         self.next_key = 0
-        self.agents = {}  # key, (task, full_message_history, model)
+        self.agents: dict[
+            int, tuple[str, list[Message], str]
+        ] = {}  # key, (task, full_message_history, model)
         self.cfg = Config()
 
     # Create new GPT agent
     # TODO: Centralise use of create_chat_completion() to globally enforce token limit
 
-    def create_agent(self, task: str, prompt: str, model: str) -> tuple[int, str]:
+    def create_agent(
+        self, task: str, creation_prompt: str, model: str
+    ) -> tuple[int, str]:
         """Create a new agent and return its key
 
         Args:
             task: The task to perform
-            prompt: The prompt to use
-            model: The model to use
+            creation_prompt: Prompt passed to the LLM at creation
+            model: The model to use to run this agent
 
         Returns:
             The key of the new agent
         """
-        messages: List[Message] = [
-            {"role": "user", "content": prompt},
-        ]
+        messages = ChatSequence.for_model(model, [Message("user", creation_prompt)])
+
         for plugin in self.cfg.plugins:
             if not plugin.can_handle_pre_instruction():
                 continue
-            if plugin_messages := plugin.pre_instruction(messages):
-                messages.extend(iter(plugin_messages))
+            if plugin_messages := plugin.pre_instruction(messages.raw()):
+                messages.extend([Message(**raw_msg) for raw_msg in plugin_messages])
         # Start GPT instance
-        agent_reply = create_chat_completion(
-            model=model,
-            messages=messages,
-        )
+        agent_reply = create_chat_completion(prompt=messages)
 
-        messages.append({"role": "assistant", "content": agent_reply})
+        messages.add("assistant", agent_reply)
 
         plugins_reply = ""
         for i, plugin in enumerate(self.cfg.plugins):
             if not plugin.can_handle_on_instruction():
                 continue
-            if plugin_result := plugin.on_instruction(messages):
+            if plugin_result := plugin.on_instruction([m.raw() for m in messages]):
                 sep = "\n" if i else ""
                 plugins_reply = f"{plugins_reply}{sep}{plugin_result}"
 
         if plugins_reply and plugins_reply != "":
-            messages.append({"role": "assistant", "content": plugins_reply})
+            messages.add("assistant", plugins_reply)
         key = self.next_key
         # This is done instead of len(agents) to make keys unique even if agents
         # are deleted
         self.next_key += 1
 
-        self.agents[key] = (task, messages, model)
+        self.agents[key] = (task, list(messages), model)
 
         for plugin in self.cfg.plugins:
             if not plugin.can_handle_post_instruction():
@@ -83,33 +82,30 @@ class AgentManager(metaclass=Singleton):
         task, messages, model = self.agents[int(key)]
 
         # Add user message to message history before sending to agent
-        messages.append({"role": "user", "content": message})
+        messages = ChatSequence.for_model(model, messages)
+        messages.add("user", message)
 
         for plugin in self.cfg.plugins:
             if not plugin.can_handle_pre_instruction():
                 continue
-            if plugin_messages := plugin.pre_instruction(messages):
-                for plugin_message in plugin_messages:
-                    messages.append(plugin_message)
+            if plugin_messages := plugin.pre_instruction([m.raw() for m in messages]):
+                messages.extend([Message(**raw_msg) for raw_msg in plugin_messages])
 
         # Start GPT instance
-        agent_reply = create_chat_completion(
-            model=model,
-            messages=messages,
-        )
+        agent_reply = create_chat_completion(prompt=messages)
 
-        messages.append({"role": "assistant", "content": agent_reply})
+        messages.add("assistant", agent_reply)
 
         plugins_reply = agent_reply
         for i, plugin in enumerate(self.cfg.plugins):
             if not plugin.can_handle_on_instruction():
                 continue
-            if plugin_result := plugin.on_instruction(messages):
+            if plugin_result := plugin.on_instruction([m.raw() for m in messages]):
                 sep = "\n" if i else ""
                 plugins_reply = f"{plugins_reply}{sep}{plugin_result}"
         # Update full message history
         if plugins_reply and plugins_reply != "":
-            messages.append({"role": "assistant", "content": plugins_reply})
+            messages.add("assistant", plugins_reply)
 
         for plugin in self.cfg.plugins:
             if not plugin.can_handle_post_instruction():

autogpt/app.py

@@ -1,21 +1,15 @@
 """ Command and Control """
 import json
-from typing import Dict, List, NoReturn, Union
+from typing import Dict, List, Union
 
+from autogpt.agent.agent import Agent
 from autogpt.agent.agent_manager import AgentManager
-from autogpt.commands.command import CommandRegistry, command
+from autogpt.commands.command import command
 from autogpt.commands.web_requests import scrape_links, scrape_text
-from autogpt.config import Config
-from autogpt.logs import logger
-from autogpt.memory import get_memory
 from autogpt.processing.text import summarize_text
-from autogpt.prompts.generator import PromptGenerator
 from autogpt.speech import say_text
 from autogpt.url_utils.validators import validate_url
 
-CFG = Config()
-AGENT_MANAGER = AgentManager()
-
 
 def is_valid_int(value: str) -> bool:
     """Check if the value is a valid integer
@@ -90,50 +84,44 @@ def map_command_synonyms(command_name: str):
 
 
 def execute_command(
-    command_registry: CommandRegistry,
     command_name: str,
-    arguments,
-    prompt: PromptGenerator,
+    arguments: dict[str, str],
+    agent: Agent,
 ):
     """Execute the command and return the result
 
     Args:
         command_name (str): The name of the command to execute
         arguments (dict): The arguments for the command
+        agent (Agent): The agent that is executing the command
 
     Returns:
         str: The result of the command
     """
     try:
-        cmd = command_registry.commands.get(command_name)
+        cmd = agent.command_registry.commands.get(command_name)
 
         # If the command is found, call it with the provided arguments
         if cmd:
-            return cmd(**arguments)
+            return cmd(**arguments, agent=agent)
 
         # TODO: Remove commands below after they are moved to the command registry.
         command_name = map_command_synonyms(command_name.lower())
 
-        if command_name == "memory_add":
-            return get_memory(CFG).add(arguments["string"])
-
         # TODO: Change these to take in a file rather than pasted code, if
         # non-file is given, return instructions "Input should be a python
         # filepath, write your code to file and try again
-        elif command_name == "task_complete":
-            shutdown()
-        else:
-            for command in prompt.commands:
-                if (
-                    command_name == command["label"].lower()
-                    or command_name == command["name"].lower()
-                ):
-                    return command["function"](**arguments)
-            return (
-                f"Unknown command '{command_name}'. Please refer to the 'COMMANDS'"
-                " list for available commands and only respond in the specified JSON"
-                " format."
-            )
+        for command in agent.ai_config.prompt_generator.commands:
+            if (
+                command_name == command["label"].lower()
+                or command_name == command["name"].lower()
+            ):
+                return command["function"](**arguments)
+        return (
+            f"Unknown command '{command_name}'. Please refer to the 'COMMANDS'"
+            " list for available commands and only respond in the specified JSON"
+            " format."
+        )
     except Exception as e:
         return f"Error: {str(e)}"
 
@@ -142,8 +130,8 @@ def execute_command(
     "get_text_summary", "Get text summary", '"url": "<url>", "question": "<question>"'
 )
 @validate_url
-def get_text_summary(url: str, question: str) -> str:
-    """Return the results of a Google search
+def get_text_summary(url: str, question: str, agent: Agent) -> str:
+    """Get the text summary of a webpage
 
     Args:
         url (str): The url to scrape
@@ -152,15 +140,16 @@ def get_text_summary(url: str, question: str) -> str:
     Returns:
         str: The summary of the text
     """
-    text = scrape_text(url)
-    summary = summarize_text(url, text, question)
+    text = scrape_text(url, agent)
+    summary, _ = summarize_text(text, question=question)
+
     return f""" "Result" : {summary}"""
 
 
-@command("get_hyperlinks", "Get text summary", '"url": "<url>"')
+@command("get_hyperlinks", "Get hyperlinks", '"url": "<url>"')
 @validate_url
-def get_hyperlinks(url: str) -> Union[str, List[str]]:
-    """Return the results of a Google search
+def get_hyperlinks(url: str, agent: Agent) -> Union[str, List[str]]:
+    """Get all hyperlinks on a webpage
 
     Args:
         url (str): The url to scrape
@@ -168,13 +157,7 @@ def get_hyperlinks(url: str) -> Union[str, List[str]]:
     Returns:
         str or list: The hyperlinks on the page
     """
-    return scrape_links(url)
-
-
-def shutdown() -> NoReturn:
-    """Shut down the program"""
-    logger.info("Shutting down...")
-    quit()
+    return scrape_links(url, agent)
 
 
 @command(
@@ -182,7 +165,7 @@ def shutdown() -> NoReturn:
     "Start GPT Agent",
     '"name": "<name>", "task": "<short_task_desc>", "prompt": "<prompt>"',
 )
-def start_agent(name: str, task: str, prompt: str, model=CFG.fast_llm_model) -> str:
+def start_agent(name: str, task: str, prompt: str, agent: Agent, model=None) -> str:
     """Start an agent with a given name, task, and prompt
 
     Args:
@@ -194,6 +177,8 @@ def start_agent(name: str, task: str, prompt: str, model=CFG.fast_llm_model) ->
     Returns:
         str: The response of the agent
     """
+    agent_manager = AgentManager()
+
     # Remove underscores from name
     voice_name = name.replace("_", " ")
 
@@ -201,48 +186,48 @@ def start_agent(name: str, task: str, prompt: str, model=CFG.fast_llm_model) ->
     agent_intro = f"{voice_name} here, Reporting for duty!"
 
     # Create agent
-    if CFG.speak_mode:
+    if agent.config.speak_mode:
         say_text(agent_intro, 1)
-    key, ack = AGENT_MANAGER.create_agent(task, first_message, model)
+    key, ack = agent_manager.create_agent(task, first_message, model)
 
-    if CFG.speak_mode:
+    if agent.config.speak_mode:
         say_text(f"Hello {voice_name}. Your task is as follows. {task}.")
 
     # Assign task (prompt), get response
-    agent_response = AGENT_MANAGER.message_agent(key, prompt)
+    agent_response = agent_manager.message_agent(key, prompt)
 
     return f"Agent {name} created with key {key}. First response: {agent_response}"
 
 
 @command("message_agent", "Message GPT Agent", '"key": "<key>", "message": "<message>"')
-def message_agent(key: str, message: str) -> str:
+def message_agent(key: str, message: str, agent: Agent) -> str:
     """Message an agent with a given key and message"""
     # Check if the key is a valid integer
     if is_valid_int(key):
-        agent_response = AGENT_MANAGER.message_agent(int(key), message)
+        agent_response = AgentManager().message_agent(int(key), message)
     else:
         return "Invalid key, must be an integer."
 
     # Speak response
-    if CFG.speak_mode:
+    if agent.config.speak_mode:
         say_text(agent_response, 1)
     return agent_response
 
 
-@command("list_agents", "List GPT Agents", "")
-def list_agents() -> str:
+@command("list_agents", "List GPT Agents", "() -> str")
+def list_agents(agent: Agent) -> str:
     """List all agents
 
     Returns:
         str: A list of all agents
     """
     return "List of agents:\n" + "\n".join(
-        [str(x[0]) + ": " + x[1] for x in AGENT_MANAGER.list_agents()]
+        [str(x[0]) + ": " + x[1] for x in AgentManager().list_agents()]
     )
 
 
 @command("delete_agent", "Delete GPT Agent", '"key": "<key>"')
-def delete_agent(key: str) -> str:
+def delete_agent(key: str, agent: Agent) -> str:
     """Delete an agent with a given key
 
     Args:
@@ -251,5 +236,5 @@ def delete_agent(key: str) -> str:
     Returns:
         str: A message indicating whether the agent was deleted or not
     """
-    result = AGENT_MANAGER.delete_agent(key)
+    result = AgentManager().delete_agent(key)
     return f"Agent {key} deleted." if result else f"Agent {key} does not exist."

autogpt/cli.py

@@ -15,6 +15,11 @@ import click
     "-C",
     help="Specifies which ai_settings.yaml file to use, will also automatically skip the re-prompt.",
 )
+@click.option(
+    "--prompt-settings",
+    "-P",
+    help="Specifies which prompt_settings.yaml file to use.",
+)
 @click.option(
     "-l",
     "--continuous-limit",
@@ -66,6 +71,7 @@ def main(
     continuous: bool,
     continuous_limit: int,
     ai_settings: str,
+    prompt_settings: str,
     skip_reprompt: bool,
     speak: bool,
     debug: bool,
@@ -91,6 +97,7 @@ def main(
             continuous,
             continuous_limit,
             ai_settings,
+            prompt_settings,
             skip_reprompt,
             speak,
             debug,

autogpt/commands/analyze_code.py

@@ -1,8 +1,9 @@
 """Code evaluation module."""
 from __future__ import annotations
 
+from autogpt.agent.agent import Agent
 from autogpt.commands.command import command
-from autogpt.llm import call_ai_function
+from autogpt.llm.utils import call_ai_function
 
 
 @command(
@@ -10,7 +11,7 @@ from autogpt.llm import call_ai_function
     "Analyze Code",
     '"code": "<full_code_string>"',
 )
-def analyze_code(code: str) -> list[str]:
+def analyze_code(code: str, agent: Agent) -> list[str]:
     """
     A function that takes in a string and returns a response from create chat
       completion api call.
@@ -28,4 +29,6 @@ def analyze_code(code: str) -> list[str]:
         "Analyzes the given code and returns a list of suggestions for improvements."
     )
 
-    return call_ai_function(function_string, args, description_string)
+    return call_ai_function(
+        function_string, args, description_string, config=agent.config
+    )

autogpt/commands/audio_text.py

@@ -3,20 +3,19 @@ import json
 
 import requests
 
+from autogpt.agent.agent import Agent
 from autogpt.commands.command import command
-from autogpt.config import Config
-
-CFG = Config()
 
 
 @command(
     "read_audio_from_file",
     "Convert Audio to text",
     '"filename": "<filename>"',
-    CFG.huggingface_audio_to_text_model,
-    "Configure huggingface_audio_to_text_model.",
+    lambda config: config.huggingface_audio_to_text_model
+    and config.huggingface_api_token,
+    "Configure huggingface_audio_to_text_model and Hugging Face api token.",
 )
-def read_audio_from_file(filename: str) -> str:
+def read_audio_from_file(filename: str, agent: Agent) -> str:
     """
     Convert audio to text.
 
@@ -28,10 +27,10 @@ def read_audio_from_file(filename: str) -> str:
     """
     with open(filename, "rb") as audio_file:
         audio = audio_file.read()
-    return read_audio(audio)
+    return read_audio(audio, agent.config)
 
 
-def read_audio(audio: bytes) -> str:
+def read_audio(audio: bytes, agent: Agent) -> str:
     """
     Convert audio to text.
 
@@ -41,9 +40,20 @@ def read_audio(audio: bytes) -> str:
     Returns:
         str: The text from the audio
     """
-    model = CFG.huggingface_audio_to_text_model
+    if agent.config.audio_to_text_provider == "huggingface":
+        text = read_huggingface_audio(audio, agent.config)
+        if text:
+            return f"The audio says: {text}"
+        else:
+            return f"Error, couldn't convert audio to text"
+
+    return "Error: No audio to text provider given"
+
+
+def read_huggingface_audio(audio: bytes, agent: Agent) -> str:
+    model = agent.config.huggingface_audio_to_text_model
     api_url = f"https://api-inference.huggingface.co/models/{model}"
-    api_token = CFG.huggingface_api_token
+    api_token = agent.config.huggingface_api_token
     headers = {"Authorization": f"Bearer {api_token}"}
 
     if api_token is None:
@@ -57,5 +67,5 @@ def read_audio(audio: bytes) -> str:
         data=audio,
     )
 
-    text = json.loads(response.content.decode("utf-8"))["text"]
-    return f"The audio says: {text}"
+    response_json = json.loads(response.content.decode("utf-8"))
+    return response_json.get("text")

autogpt/commands/command.py

@@ -1,8 +1,12 @@
 import functools
 import importlib
 import inspect
+from inspect import Parameter
 from typing import Any, Callable, Optional
 
+from autogpt.config import Config
+from autogpt.logs import logger
+
 # Unique identifier for auto-gpt commands
 AUTO_GPT_COMMAND_IDENTIFIER = "auto_gpt_command"
 
@@ -22,19 +26,23 @@ class Command:
         description: str,
         method: Callable[..., Any],
         signature: str = "",
-        enabled: bool = True,
+        enabled: bool | Callable[[Config], bool] = True,
         disabled_reason: Optional[str] = None,
     ):
         self.name = name
         self.description = description
         self.method = method
-        self.signature = signature if signature else str(inspect.signature(self.method))
+        self.signature = signature
         self.enabled = enabled
         self.disabled_reason = disabled_reason
 
     def __call__(self, *args, **kwargs) -> Any:
+        if hasattr(kwargs, "config") and callable(self.enabled):
+            self.enabled = self.enabled(kwargs["config"])
         if not self.enabled:
-            return f"Command '{self.name}' is disabled: {self.disabled_reason}"
+            if self.disabled_reason:
+                return f"Command '{self.name}' is disabled: {self.disabled_reason}"
+            return f"Command '{self.name}' is disabled"
         return self.method(*args, **kwargs)
 
     def __str__(self) -> str:
@@ -59,6 +67,10 @@ class CommandRegistry:
         return importlib.reload(module)
 
     def register(self, cmd: Command) -> None:
+        if cmd.name in self.commands:
+            logger.warn(
+                f"Command '{cmd.name}' already registered and will be overwritten!"
+            )
         self.commands[cmd.name] = cmd
 
     def unregister(self, command_name: str):
@@ -127,12 +139,22 @@ class CommandRegistry:
 def command(
     name: str,
     description: str,
-    signature: str = "",
-    enabled: bool = True,
+    signature: str,
+    enabled: bool | Callable[[Config], bool] = True,
     disabled_reason: Optional[str] = None,
 ) -> Callable[..., Any]:
     """The command decorator is used to create Command objects from ordinary functions."""
 
+    # TODO: Remove this in favor of better command management
+    CFG = Config()
+
+    if callable(enabled):
+        enabled = enabled(CFG)
+    if not enabled:
+        if disabled_reason is not None:
+            logger.debug(f"Command '{name}' is disabled: {disabled_reason}")
+        return lambda func: func
+
     def decorator(func: Callable[..., Any]) -> Command:
         cmd = Command(
             name=name,
@@ -154,3 +176,32 @@ def command(
         return wrapper
 
     return decorator
+
+
+def ignore_unexpected_kwargs(func: Callable[..., Any]) -> Callable[..., Any]:
+    def filter_kwargs(kwargs: dict) -> dict:
+        sig = inspect.signature(func)
+        # Parameter.VAR_KEYWORD - a dict of keyword arguments that aren't bound to any other
+        if any(map(lambda p: p.kind == Parameter.VAR_KEYWORD, sig.parameters.values())):
+            # if **kwargs exist, return directly
+            return kwargs
+
+        _params = list(
+            filter(
+                lambda p: p.kind
+                in {Parameter.KEYWORD_ONLY, Parameter.POSITIONAL_OR_KEYWORD},
+                sig.parameters.values(),
+            )
+        )
+
+        res_kwargs = {
+            param.name: kwargs[param.name] for param in _params if param.name in kwargs
+        }
+        return res_kwargs
+
+    @functools.wraps(func)
+    def wrapper(*args, **kwargs) -> Any:
+        kwargs = filter_kwargs(kwargs)
+        return func(*args, **kwargs)
+
+    return wrapper

autogpt/commands/execute_code.py

@@ -6,15 +6,53 @@ from pathlib import Path
 import docker
 from docker.errors import ImageNotFound
 
+from autogpt.agent.agent import Agent
 from autogpt.commands.command import command
 from autogpt.config import Config
 from autogpt.logs import logger
+from autogpt.setup import CFG
+from autogpt.workspace.workspace import Workspace
 
-CFG = Config()
+ALLOWLIST_CONTROL = "allowlist"
+DENYLIST_CONTROL = "denylist"
+
+
+@command(
+    "execute_python_code",
+    "Create a Python file and execute it",
+    '"code": "<code>", "basename": "<basename>"',
+)
+def execute_python_code(code: str, basename: str, agent: Agent) -> str:
+    """Create and execute a Python file in a Docker container and return the STDOUT of the
+    executed code. If there is any data that needs to be captured use a print statement
+
+    Args:
+        code (str): The Python code to run
+        basename (str): A name to be given to the Python file
+
+    Returns:
+        str: The STDOUT captured from the code when it ran
+    """
+    ai_name = agent.ai_name
+    directory = os.path.join(agent.config.workspace_path, ai_name, "executed_code")
+    os.makedirs(directory, exist_ok=True)
+
+    if not basename.endswith(".py"):
+        basename = basename + ".py"
+
+    path = os.path.join(directory, basename)
+
+    try:
+        with open(path, "w+", encoding="utf-8") as f:
+            f.write(code)
+
+        return execute_python_file(f.name, agent)
+    except Exception as e:
+        return f"Error: {str(e)}"
 
 
 @command("execute_python_file", "Execute Python File", '"filename": "<filename>"')
-def execute_python_file(filename: str) -> str:
+def execute_python_file(filename: str, agent: Agent) -> str:
     """Execute a Python file in a Docker container and return the output
 
     Args:
@@ -23,17 +61,30 @@ def execute_python_file(filename: str) -> str:
     Returns:
         str: The output of the file
     """
-    logger.info(f"Executing file '{filename}'")
+    logger.info(
+        f"Executing python file '{filename}' in working directory '{CFG.workspace_path}'"
+    )
 
     if not filename.endswith(".py"):
         return "Error: Invalid file type. Only .py files are allowed."
 
-    if not os.path.isfile(filename):
-        return f"Error: File '{filename}' does not exist."
+    workspace = Workspace(
+        agent.config.workspace_path, agent.config.restrict_to_workspace
+    )
+
+    path = workspace.get_path(filename)
+    if not path.is_file():
+        # Mimic the response that you get from the command line so that it's easier to identify
+        return (
+            f"python: can't open file '{filename}': [Errno 2] No such file or directory"
+        )
 
     if we_are_running_in_a_docker_container():
         result = subprocess.run(
-            f"python {filename}", capture_output=True, encoding="utf8", shell=True
+            ["python", str(path)],
+            capture_output=True,
+            encoding="utf8",
+            cwd=CFG.workspace_path,
         )
         if result.returncode == 0:
             return result.stdout
@@ -65,9 +116,9 @@ def execute_python_file(filename: str) -> str:
                     logger.info(status)
         container = client.containers.run(
             image_name,
-            f"python {Path(filename).relative_to(CFG.workspace_path)}",
+            ["python", str(path.relative_to(workspace.root))],
             volumes={
-                CFG.workspace_path: {
+                agent.config.workspace_path: {
                     "bind": "/workspace",
                     "mode": "ro",
                 }
@@ -97,16 +148,36 @@ def execute_python_file(filename: str) -> str:
         return f"Error: {str(e)}"
 
 
+def validate_command(command: str, config: Config) -> bool:
+    """Validate a command to ensure it is allowed
+
+    Args:
+        command (str): The command to validate
+
+    Returns:
+        bool: True if the command is allowed, False otherwise
+    """
+    if not command:
+        return False
+
+    command_name = command.split()[0]
+
+    if config.shell_command_control == ALLOWLIST_CONTROL:
+        return command_name in config.shell_allowlist
+    else:
+        return command_name not in config.shell_denylist
+
+
 @command(
     "execute_shell",
     "Execute Shell Command, non-interactive commands only",
     '"command_line": "<command_line>"',
-    CFG.execute_local_commands,
+    lambda cfg: cfg.execute_local_commands,
     "You are not allowed to run local shell commands. To execute"
     " shell commands, EXECUTE_LOCAL_COMMANDS must be set to 'True' "
-    "in your config. Do not attempt to bypass the restriction.",
+    "in your config file: .env - do not attempt to bypass the restriction.",
 )
-def execute_shell(command_line: str) -> str:
+def execute_shell(command_line: str, agent: Agent) -> str:
     """Execute a shell command and return the output
 
     Args:
@@ -115,11 +186,14 @@ def execute_shell(command_line: str) -> str:
     Returns:
         str: The output of the command
     """
+    if not validate_command(command_line, agent.config):
+        logger.info(f"Command '{command_line}' not allowed")
+        return "Error: This Shell Command is not allowed."
 
     current_dir = Path.cwd()
     # Change dir into workspace if necessary
-    if not current_dir.is_relative_to(CFG.workspace_path):
-        os.chdir(CFG.workspace_path)
+    if not current_dir.is_relative_to(agent.config.workspace_path):
+        os.chdir(agent.config.workspace_path)
 
     logger.info(
         f"Executing command '{command_line}' in working directory '{os.getcwd()}'"
@@ -138,12 +212,12 @@ def execute_shell(command_line: str) -> str:
     "execute_shell_popen",
     "Execute Shell Command, non-interactive commands only",
     '"command_line": "<command_line>"',
-    CFG.execute_local_commands,
+    lambda config: config.execute_local_commands,
     "You are not allowed to run local shell commands. To execute"
     " shell commands, EXECUTE_LOCAL_COMMANDS must be set to 'True' "
     "in your config. Do not attempt to bypass the restriction.",
 )
-def execute_shell_popen(command_line) -> str:
+def execute_shell_popen(command_line, agent: Agent) -> str:
     """Execute a shell command with Popen and returns an english description
     of the event and the process id
 
@@ -153,11 +227,14 @@ def execute_shell_popen(command_line) -> str:
     Returns:
         str: Description of the fact that the process started and its id
     """
+    if not validate_command(command_line, agent.config):
+        logger.info(f"Command '{command_line}' not allowed")
+        return "Error: This Shell Command is not allowed."
 
     current_dir = os.getcwd()
     # Change dir into workspace if necessary
-    if CFG.workspace_path not in current_dir:
-        os.chdir(CFG.workspace_path)
+    if agent.config.workspace_path not in current_dir:
+        os.chdir(agent.config.workspace_path)
 
     logger.info(
         f"Executing command '{command_line}' in working directory '{os.getcwd()}'"

autogpt/commands/file_operations.py

@@ -1,83 +1,127 @@
 """File operations for AutoGPT"""
 from __future__ import annotations
 
+import hashlib
 import os
 import os.path
-from typing import Generator
+import re
+from typing import Generator, Literal
 
 import requests
 from colorama import Back, Fore
+from confection import Config
 from requests.adapters import HTTPAdapter, Retry
 
-from autogpt.commands.command import command
+from autogpt.agent.agent import Agent
+from autogpt.commands.command import command, ignore_unexpected_kwargs
+from autogpt.commands.file_operations_utils import read_textual_file
 from autogpt.config import Config
 from autogpt.logs import logger
+from autogpt.memory.vector import MemoryItem, VectorMemory
 from autogpt.spinner import Spinner
 from autogpt.utils import readable_file_size
 
-CFG = Config()
+Operation = Literal["write", "append", "delete"]
 
 
-def check_duplicate_operation(operation: str, filename: str) -> bool:
-    """Check if the operation has already been performed on the given file
+def text_checksum(text: str) -> str:
+    """Get the hex checksum for the given text."""
+    return hashlib.md5(text.encode("utf-8")).hexdigest()
 
-    Args:
-        operation (str): The operation to check for
-        filename (str): The name of the file to check for
+
+def operations_from_log(
+    log_path: str,
+) -> Generator[tuple[Operation, str, str | None], None, None]:
+    """Parse the file operations log and return a tuple containing the log entries"""
+    try:
+        log = open(log_path, "r", encoding="utf-8")
+    except FileNotFoundError:
+        return
+
+    for line in log:
+        line = line.replace("File Operation Logger", "").strip()
+        if not line:
+            continue
+        operation, tail = line.split(": ", maxsplit=1)
+        operation = operation.strip()
+        if operation in ("write", "append"):
+            try:
+                path, checksum = (x.strip() for x in tail.rsplit(" #", maxsplit=1))
+            except ValueError:
+                logger.warn(f"File log entry lacks checksum: '{line}'")
+                path, checksum = tail.strip(), None
+            yield (operation, path, checksum)
+        elif operation == "delete":
+            yield (operation, tail.strip(), None)
+
+    log.close()
+
+
+def file_operations_state(log_path: str) -> dict[str, str]:
+    """Iterates over the operations log and returns the expected state.
+
+    Parses a log file at config.file_logger_path to construct a dictionary that maps
+    each file path written or appended to its checksum. Deleted files are removed
+    from the dictionary.
 
     Returns:
-        bool: True if the operation has already been performed on the file
+        A dictionary mapping file paths to their checksums.
+
+    Raises:
+        FileNotFoundError: If config.file_logger_path is not found.
+        ValueError: If the log file content is not in the expected format.
     """
-    log_content = read_file(CFG.file_logger_path)
-    log_entry = f"{operation}: {filename}\n"
-    return log_entry in log_content
+    state = {}
+    for operation, path, checksum in operations_from_log(log_path):
+        if operation in ("write", "append"):
+            state[path] = checksum
+        elif operation == "delete":
+            del state[path]
+    return state
 
 
-def log_operation(operation: str, filename: str) -> None:
+def is_duplicate_operation(
+    operation: Operation, filename: str, config: Config, checksum: str | None = None
+) -> bool:
+    """Check if the operation has already been performed
+
+    Args:
+        operation: The operation to check for
+        filename: The name of the file to check for
+        checksum: The checksum of the contents to be written
+
+    Returns:
+        True if the operation has already been performed on the file
+    """
+    state = file_operations_state(config.file_logger_path)
+    if operation == "delete" and filename not in state:
+        return True
+    if operation == "write" and state.get(filename) == checksum:
+        return True
+    return False
+
+
+def log_operation(
+    operation: str, filename: str, agent: Agent, checksum: str | None = None
+) -> None:
     """Log the file operation to the file_logger.txt
 
     Args:
-        operation (str): The operation to log
-        filename (str): The name of the file the operation was performed on
+        operation: The operation to log
+        filename: The name of the file the operation was performed on
+        checksum: The checksum of the contents to be written
     """
-    log_entry = f"{operation}: {filename}\n"
-    append_to_file(CFG.file_logger_path, log_entry, should_log=False)
+    log_entry = f"{operation}: {filename}"
+    if checksum is not None:
+        log_entry += f" #{checksum}"
+    logger.debug(f"Logging file operation: {log_entry}")
+    append_to_file(
+        agent.config.file_logger_path, f"{log_entry}\n", agent, should_log=False
+    )
 
 
-def split_file(
-    content: str, max_length: int = 4000, overlap: int = 0
-) -> Generator[str, None, None]:
-    """
-    Split text into chunks of a specified maximum length with a specified overlap
-    between chunks.
-
-    :param content: The input text to be split into chunks
-    :param max_length: The maximum length of each chunk,
-        default is 4000 (about 1k token)
-    :param overlap: The number of overlapping characters between chunks,
-        default is no overlap
-    :return: A generator yielding chunks of text
-    """
-    start = 0
-    content_length = len(content)
-
-    while start < content_length:
-        end = start + max_length
-        if end + overlap < content_length:
-            chunk = content[start : end + overlap - 1]
-        else:
-            chunk = content[start:content_length]
-
-            # Account for the case where the last chunk is shorter than the overlap, so it has already been consumed
-            if len(chunk) <= overlap:
-                break
-
-        yield chunk
-        start += max_length - overlap
-
-
-@command("read_file", "Read file", '"filename": "<filename>"')
-def read_file(filename: str) -> str:
+@command("read_file", "Read a file", '"filename": "<filename>"')
+def read_file(filename: str, agent: Agent) -> str:
     """Read a file and return the contents
 
     Args:
@@ -87,49 +131,46 @@ def read_file(filename: str) -> str:
         str: The contents of the file
     """
     try:
-        with open(filename, "r", encoding="utf-8") as f:
-            content = f.read()
+        content = read_textual_file(filename, logger)
+
+        # TODO: invalidate/update memory when file is edited
+        file_memory = MemoryItem.from_text_file(content, filename)
+        if len(file_memory.chunks) > 1:
+            return file_memory.summary
+
         return content
     except Exception as e:
         return f"Error: {str(e)}"
 
 
 def ingest_file(
-    filename: str, memory, max_length: int = 4000, overlap: int = 200
+    filename: str,
+    memory: VectorMemory,
 ) -> None:
     """
     Ingest a file by reading its content, splitting it into chunks with a specified
     maximum length and overlap, and adding the chunks to the memory storage.
 
-    :param filename: The name of the file to ingest
-    :param memory: An object with an add() method to store the chunks in memory
-    :param max_length: The maximum length of each chunk, default is 4000
-    :param overlap: The number of overlapping characters between chunks, default is 200
+    Args:
+        filename: The name of the file to ingest
+        memory: An object with an add() method to store the chunks in memory
     """
     try:
-        logger.info(f"Working with file {filename}")
+        logger.info(f"Ingesting file {filename}")
         content = read_file(filename)
-        content_length = len(content)
-        logger.info(f"File length: {content_length} characters")
 
-        chunks = list(split_file(content, max_length=max_length, overlap=overlap))
+        # TODO: differentiate between different types of files
+        file_memory = MemoryItem.from_text_file(content, filename)
+        logger.debug(f"Created memory: {file_memory.dump()}")
+        memory.add(file_memory)
 
-        num_chunks = len(chunks)
-        for i, chunk in enumerate(chunks):
-            logger.info(f"Ingesting chunk {i + 1} / {num_chunks} into memory")
-            memory_to_add = (
-                f"Filename: {filename}\n" f"Content part#{i + 1}/{num_chunks}: {chunk}"
-            )
-
-            memory.add(memory_to_add)
-
-        logger.info(f"Done ingesting {num_chunks} chunks from {filename}.")
-    except Exception as e:
-        logger.info(f"Error while ingesting file '{filename}': {str(e)}")
+        logger.info(f"Ingested {len(file_memory.e_chunks)} chunks from {filename}")
+    except Exception as err:
+        logger.warn(f"Error while ingesting file '{filename}': {err}")
 
 
 @command("write_to_file", "Write to file", '"filename": "<filename>", "text": "<text>"')
-def write_to_file(filename: str, text: str) -> str:
+def write_to_file(filename: str, text: str, agent: Agent) -> str:
     """Write text to a file
 
     Args:
@@ -139,23 +180,88 @@ def write_to_file(filename: str, text: str) -> str:
     Returns:
         str: A message indicating success or failure
     """
-    if check_duplicate_operation("write", filename):
+    checksum = text_checksum(text)
+    if is_duplicate_operation("write", filename, agent.config, checksum):
         return "Error: File has already been updated."
     try:
         directory = os.path.dirname(filename)
         os.makedirs(directory, exist_ok=True)
         with open(filename, "w", encoding="utf-8") as f:
             f.write(text)
-        log_operation("write", filename)
+        log_operation("write", filename, agent, checksum)
         return "File written to successfully."
+    except Exception as err:
+        return f"Error: {err}"
+
+
+@command(
+    "replace_in_file",
+    "Replace text or code in a file",
+    '"filename": "<filename>", '
+    '"old_text": "<old_text>", "new_text": "<new_text>", '
+    '"occurrence_index": "<occurrence_index>"',
+)
+def replace_in_file(
+    filename: str, old_text: str, new_text: str, agent: Agent, occurrence_index=None
+):
+    """Update a file by replacing one or all occurrences of old_text with new_text using Python's built-in string
+    manipulation and regular expression modules for cross-platform file editing similar to sed and awk.
+
+    Args:
+        filename (str): The name of the file
+        old_text (str): String to be replaced. \n will be stripped from the end.
+        new_text (str): New string. \n will be stripped from the end.
+        occurrence_index (int): Optional index of the occurrence to replace. If None, all occurrences will be replaced.
+
+    Returns:
+        str: A message indicating whether the file was updated successfully or if there were no matches found for old_text
+        in the file.
+
+    Raises:
+        Exception: If there was an error updating the file.
+    """
+    try:
+        with open(filename, "r", encoding="utf-8") as f:
+            content = f.read()
+
+        old_text = old_text.rstrip("\n")
+        new_text = new_text.rstrip("\n")
+
+        if occurrence_index is None:
+            new_content = content.replace(old_text, new_text)
+        else:
+            matches = list(re.finditer(re.escape(old_text), content))
+            if not matches:
+                return f"No matches found for {old_text} in {filename}"
+
+            if int(occurrence_index) >= len(matches):
+                return f"Occurrence index {occurrence_index} is out of range for {old_text} in {filename}"
+
+            match = matches[int(occurrence_index)]
+            start, end = match.start(), match.end()
+            new_content = content[:start] + new_text + content[end:]
+
+        if content == new_content:
+            return f"No matches found for {old_text} in {filename}"
+
+        with open(filename, "w", encoding="utf-8") as f:
+            f.write(new_content)
+
+        with open(filename, "r", encoding="utf-8") as f:
+            checksum = text_checksum(f.read())
+        log_operation("update", filename, agent, checksum=checksum)
+
+        return f"File {filename} updated successfully."
     except Exception as e:
-        return f"Error: {str(e)}"
+        return "Error: " + str(e)
 
 
 @command(
     "append_to_file", "Append to file", '"filename": "<filename>", "text": "<text>"'
 )
-def append_to_file(filename: str, text: str, should_log: bool = True) -> str:
+def append_to_file(
+    filename: str, text: str, agent: Agent, should_log: bool = True
+) -> str:
     """Append text to a file
 
     Args:
@@ -169,19 +275,21 @@ def append_to_file(filename: str, text: str, should_log: bool = True) -> str:
     try:
         directory = os.path.dirname(filename)
         os.makedirs(directory, exist_ok=True)
-        with open(filename, "a") as f:
+        with open(filename, "a", encoding="utf-8") as f:
             f.write(text)
 
         if should_log:
-            log_operation("append", filename)
+            with open(filename, "r", encoding="utf-8") as f:
+                checksum = text_checksum(f.read())
+            log_operation("append", filename, agent, checksum=checksum)
 
         return "Text appended successfully."
-    except Exception as e:
-        return f"Error: {str(e)}"
+    except Exception as err:
+        return f"Error: {err}"
 
 
 @command("delete_file", "Delete file", '"filename": "<filename>"')
-def delete_file(filename: str) -> str:
+def delete_file(filename: str, agent: Agent) -> str:
     """Delete a file
 
     Args:
@@ -190,19 +298,20 @@ def delete_file(filename: str) -> str:
     Returns:
         str: A message indicating success or failure
     """
-    if check_duplicate_operation("delete", filename):
+    if is_duplicate_operation("delete", filename, agent.config):
         return "Error: File has already been deleted."
     try:
         os.remove(filename)
-        log_operation("delete", filename)
+        log_operation("delete", filename, agent)
         return "File deleted successfully."
-    except Exception as e:
-        return f"Error: {str(e)}"
+    except Exception as err:
+        return f"Error: {err}"
 
 
-@command("search_files", "Search Files", '"directory": "<directory>"')
-def search_files(directory: str) -> list[str]:
-    """Search for files in a directory
+@command("list_files", "List Files in Directory", '"directory": "<directory>"')
+@ignore_unexpected_kwargs
+def list_files(directory: str, agent: Agent) -> list[str]:
+    """lists files in a directory recursively
 
     Args:
         directory (str): The directory to search in
@@ -217,7 +326,7 @@ def search_files(directory: str) -> list[str]:
             if file.startswith("."):
                 continue
             relative_path = os.path.relpath(
-                os.path.join(root, file), CFG.workspace_path
+                os.path.join(root, file), agent.config.workspace_path
             )
             found_files.append(relative_path)
 
@@ -228,10 +337,10 @@ def search_files(directory: str) -> list[str]:
     "download_file",
     "Download File",
     '"url": "<url>", "filename": "<filename>"',
-    CFG.allow_downloads,
+    lambda config: config.allow_downloads,
     "Error: You do not have user authorization to download files locally.",
 )
-def download_file(url, filename):
+def download_file(url, filename, agent: Agent):
     """Downloads a file
     Args:
         url (str): URL of the file to download
@@ -241,7 +350,7 @@ def download_file(url, filename):
         directory = os.path.dirname(filename)
         os.makedirs(directory, exist_ok=True)
         message = f"{Fore.YELLOW}Downloading file from {Back.LIGHTBLUE_EX}{url}{Back.RESET}{Fore.RESET}"
-        with Spinner(message) as spinner:
+        with Spinner(message, plain_output=agent.config.plain_output) as spinner:
             session = requests.Session()
             retry = Retry(total=3, backoff_factor=1, status_forcelist=[502, 503, 504])
             adapter = HTTPAdapter(max_retries=retry)
@@ -266,7 +375,7 @@ def download_file(url, filename):
                         spinner.update_message(f"{message} {progress}")
 
             return f'Successfully downloaded and locally stored file: "{filename}"! (Size: {readable_file_size(downloaded_size)})'
-    except requests.HTTPError as e:
-        return f"Got an HTTP Error whilst trying to download file: {e}"
-    except Exception as e:
-        return "Error: " + str(e)
+    except requests.HTTPError as err:
+        return f"Got an HTTP Error whilst trying to download file: {err}"
+    except Exception as err:
+        return f"Error: {err}"

autogpt/commands/file_operations_utils.py

@@ -0,0 +1,161 @@
+import json
+import os
+
+import charset_normalizer
+import docx
+import markdown
+import PyPDF2
+import yaml
+from bs4 import BeautifulSoup
+from pylatexenc.latex2text import LatexNodes2Text
+
+from autogpt import logs
+from autogpt.logs import logger
+
+
+class ParserStrategy:
+    def read(self, file_path: str) -> str:
+        raise NotImplementedError
+
+
+# Basic text file reading
+class TXTParser(ParserStrategy):
+    def read(self, file_path: str) -> str:
+        charset_match = charset_normalizer.from_path(file_path).best()
+        logger.debug(f"Reading '{file_path}' with encoding '{charset_match.encoding}'")
+        return str(charset_match)
+
+
+# Reading text from binary file using pdf parser
+class PDFParser(ParserStrategy):
+    def read(self, file_path: str) -> str:
+        parser = PyPDF2.PdfReader(file_path)
+        text = ""
+        for page_idx in range(len(parser.pages)):
+            text += parser.pages[page_idx].extract_text()
+        return text
+
+
+# Reading text from binary file using docs parser
+class DOCXParser(ParserStrategy):
+    def read(self, file_path: str) -> str:
+        doc_file = docx.Document(file_path)
+        text = ""
+        for para in doc_file.paragraphs:
+            text += para.text
+        return text
+
+
+# Reading as dictionary and returning string format
+class JSONParser(ParserStrategy):
+    def read(self, file_path: str) -> str:
+        with open(file_path, "r") as f:
+            data = json.load(f)
+            text = str(data)
+        return text
+
+
+class XMLParser(ParserStrategy):
+    def read(self, file_path: str) -> str:
+        with open(file_path, "r") as f:
+            soup = BeautifulSoup(f, "xml")
+            text = soup.get_text()
+        return text
+
+
+# Reading as dictionary and returning string format
+class YAMLParser(ParserStrategy):
+    def read(self, file_path: str) -> str:
+        with open(file_path, "r") as f:
+            data = yaml.load(f, Loader=yaml.FullLoader)
+            text = str(data)
+        return text
+
+
+class HTMLParser(ParserStrategy):
+    def read(self, file_path: str) -> str:
+        with open(file_path, "r") as f:
+            soup = BeautifulSoup(f, "html.parser")
+            text = soup.get_text()
+        return text
+
+
+class MarkdownParser(ParserStrategy):
+    def read(self, file_path: str) -> str:
+        with open(file_path, "r") as f:
+            html = markdown.markdown(f.read())
+            text = "".join(BeautifulSoup(html, "html.parser").findAll(string=True))
+        return text
+
+
+class LaTeXParser(ParserStrategy):
+    def read(self, file_path: str) -> str:
+        with open(file_path, "r") as f:
+            latex = f.read()
+        text = LatexNodes2Text().latex_to_text(latex)
+        return text
+
+
+class FileContext:
+    def __init__(self, parser: ParserStrategy, logger: logs.Logger):
+        self.parser = parser
+        self.logger = logger
+
+    def set_parser(self, parser: ParserStrategy) -> None:
+        self.logger.debug(f"Setting Context Parser to {parser}")
+        self.parser = parser
+
+    def read_file(self, file_path) -> str:
+        self.logger.debug(f"Reading file {file_path} with parser {self.parser}")
+        return self.parser.read(file_path)
+
+
+extension_to_parser = {
+    ".txt": TXTParser(),
+    ".csv": TXTParser(),
+    ".pdf": PDFParser(),
+    ".docx": DOCXParser(),
+    ".json": JSONParser(),
+    ".xml": XMLParser(),
+    ".yaml": YAMLParser(),
+    ".yml": YAMLParser(),
+    ".html": HTMLParser(),
+    ".htm": HTMLParser(),
+    ".xhtml": HTMLParser(),
+    ".md": MarkdownParser(),
+    ".markdown": MarkdownParser(),
+    ".tex": LaTeXParser(),
+}
+
+
+def is_file_binary_fn(file_path: str):
+    """Given a file path load all its content and checks if the null bytes is present
+
+    Args:
+        file_path (_type_): _description_
+
+    Returns:
+        bool: is_binary
+    """
+    with open(file_path, "rb") as f:
+        file_data = f.read()
+    if b"\x00" in file_data:
+        return True
+    return False
+
+
+def read_textual_file(file_path: str, logger: logs.Logger) -> str:
+    if not os.path.isfile(file_path):
+        raise FileNotFoundError(
+            f"read_file {file_path} failed: no such file or directory"
+        )
+    is_binary = is_file_binary_fn(file_path)
+    file_extension = os.path.splitext(file_path)[1].lower()
+    parser = extension_to_parser.get(file_extension)
+    if not parser:
+        if is_binary:
+            raise ValueError(f"Unsupported binary file format: {file_extension}")
+        # fallback to txt file parser (to support script and code files loading)
+        parser = TXTParser()
+    file_context = FileContext(parser, logger)
+    return file_context.read_file(file_path)

autogpt/commands/git_operations.py

@@ -1,22 +1,21 @@
 """Git operations for autogpt"""
+
 from git.repo import Repo
 
+from autogpt.agent.agent import Agent
 from autogpt.commands.command import command
-from autogpt.config import Config
 from autogpt.url_utils.validators import validate_url
 
-CFG = Config()
-
 
 @command(
     "clone_repository",
     "Clone Repository",
     '"url": "<repository_url>", "clone_path": "<clone_path>"',
-    CFG.github_username and CFG.github_api_key,
+    lambda config: config.github_username and config.github_api_key,
     "Configure github_username and github_api_key.",
 )
 @validate_url
-def clone_repository(url: str, clone_path: str) -> str:
+def clone_repository(url: str, clone_path: str, agent: Agent) -> str:
     """Clone a GitHub repository locally.
 
     Args:
@@ -27,7 +26,11 @@ def clone_repository(url: str, clone_path: str) -> str:
         str: The result of the clone operation.
     """
     split_url = url.split("//")
-    auth_repo_url = f"//{CFG.github_username}:{CFG.github_api_key}@".join(split_url)
+    auth_repo_url = (
+        f"//{agent.config.github_username}:{agent.config.github_api_key}@".join(
+            split_url
+        )
+    )
     try:
         Repo.clone_from(url=auth_repo_url, to_path=clone_path)
         return f"""Cloned {url} to {clone_path}"""

autogpt/commands/google_search.py

@@ -2,17 +2,24 @@
 from __future__ import annotations
 
 import json
+import time
+from itertools import islice
 
-from duckduckgo_search import ddg
+from duckduckgo_search import DDGS
 
+from autogpt.agent.agent import Agent
 from autogpt.commands.command import command
-from autogpt.config import Config
 
-CFG = Config()
+DUCKDUCKGO_MAX_ATTEMPTS = 3
 
 
-@command("google", "Google Search", '"query": "<query>"', not CFG.google_api_key)
-def google_search(query: str, num_results: int = 8) -> str:
+@command(
+    "google",
+    "Google Search",
+    '"query": "<query>"',
+    lambda config: not config.google_api_key,
+)
+def google_search(query: str, agent: Agent, num_results: int = 8) -> str:
     """Return the results of a Google search
 
     Args:
@@ -23,15 +30,20 @@ def google_search(query: str, num_results: int = 8) -> str:
         str: The results of the search.
     """
     search_results = []
-    if not query:
-        return json.dumps(search_results)
+    attempts = 0
 
-    results = ddg(query, max_results=num_results)
-    if not results:
-        return json.dumps(search_results)
+    while attempts < DUCKDUCKGO_MAX_ATTEMPTS:
+        if not query:
+            return json.dumps(search_results)
 
-    for j in results:
-        search_results.append(j)
+        results = DDGS().text(query)
+        search_results = list(islice(results, num_results))
+
+        if search_results:
+            break
+
+        time.sleep(1)
+        attempts += 1
 
     results = json.dumps(search_results, ensure_ascii=False, indent=4)
     return safe_google_results(results)
@@ -41,10 +53,13 @@ def google_search(query: str, num_results: int = 8) -> str:
     "google",
     "Google Search",
     '"query": "<query>"',
-    bool(CFG.google_api_key),
-    "Configure google_api_key.",
+    lambda config: bool(config.google_api_key)
+    and bool(config.google_custom_search_engine_id),
+    "Configure google_api_key and custom_search_engine_id.",
 )
-def google_official_search(query: str, num_results: int = 8) -> str | list[str]:
+def google_official_search(
+    query: str, agent: Agent, num_results: int = 8
+) -> str | list[str]:
     """Return the results of a Google search using the official Google API
 
     Args:
@@ -60,8 +75,8 @@ def google_official_search(query: str, num_results: int = 8) -> str | list[str]:
 
     try:
         # Get the Google API key and Custom Search Engine ID from the config file
-        api_key = CFG.google_api_key
-        custom_search_engine_id = CFG.custom_search_engine_id
+        api_key = agent.config.google_api_key
+        custom_search_engine_id = agent.config.google_custom_search_engine_id
 
         # Initialize the Custom Search API service
         service = build("customsearch", "v1", developerKey=api_key)
@@ -110,7 +125,7 @@ def safe_google_results(results: str | list) -> str:
     """
     if isinstance(results, list):
         safe_message = json.dumps(
-            [result.encode("utf-8", "ignore") for result in results]
+            [result.encode("utf-8", "ignore").decode("utf-8") for result in results]
         )
     else:
         safe_message = results.encode("utf-8", "ignore").decode("utf-8")

autogpt/commands/image_gen.py

@@ -1,5 +1,7 @@
 """ Image Generation Module for AutoGPT."""
 import io
+import json
+import time
 import uuid
 from base64 import b64decode
 
@@ -7,15 +9,19 @@ import openai
 import requests
 from PIL import Image
 
+from autogpt.agent.agent import Agent
 from autogpt.commands.command import command
-from autogpt.config import Config
 from autogpt.logs import logger
 
-CFG = Config()
 
-
-@command("generate_image", "Generate Image", '"prompt": "<prompt>"', CFG.image_provider)
-def generate_image(prompt: str, size: int = 256) -> str:
+@command(
+    "generate_image",
+    "Generate Image",
+    '"prompt": "<prompt>"',
+    lambda config: config.image_provider,
+    "Requires a image provider to be set.",
+)
+def generate_image(prompt: str, agent: Agent, size: int = 256) -> str:
     """Generate an image from a prompt.
 
     Args:
@@ -25,21 +31,21 @@ def generate_image(prompt: str, size: int = 256) -> str:
     Returns:
         str: The filename of the image
     """
-    filename = f"{CFG.workspace_path}/{str(uuid.uuid4())}.jpg"
+    filename = f"{agent.config.workspace_path}/{str(uuid.uuid4())}.jpg"
 
     # DALL-E
-    if CFG.image_provider == "dalle":
-        return generate_image_with_dalle(prompt, filename, size)
+    if agent.config.image_provider == "dalle":
+        return generate_image_with_dalle(prompt, filename, size, agent)
     # HuggingFace
-    elif CFG.image_provider == "huggingface":
-        return generate_image_with_hf(prompt, filename)
+    elif agent.config.image_provider == "huggingface":
+        return generate_image_with_hf(prompt, filename, agent)
     # SD WebUI
-    elif CFG.image_provider == "sdwebui":
-        return generate_image_with_sd_webui(prompt, filename, size)
+    elif agent.config.image_provider == "sdwebui":
+        return generate_image_with_sd_webui(prompt, filename, agent, size)
     return "No Image Provider Set"
 
 
-def generate_image_with_hf(prompt: str, filename: str) -> str:
+def generate_image_with_hf(prompt: str, filename: str, agent: Agent) -> str:
     """Generate an image with HuggingFace's API.
 
     Args:
@@ -49,35 +55,57 @@ def generate_image_with_hf(prompt: str, filename: str) -> str:
     Returns:
         str: The filename of the image
     """
-    API_URL = (
-        f"https://api-inference.huggingface.co/models/{CFG.huggingface_image_model}"
-    )
-    if CFG.huggingface_api_token is None:
+    API_URL = f"https://api-inference.huggingface.co/models/{agent.config.huggingface_image_model}"
+    if agent.config.huggingface_api_token is None:
         raise ValueError(
             "You need to set your Hugging Face API token in the config file."
         )
     headers = {
-        "Authorization": f"Bearer {CFG.huggingface_api_token}",
+        "Authorization": f"Bearer {agent.config.huggingface_api_token}",
         "X-Use-Cache": "false",
     }
 
-    response = requests.post(
-        API_URL,
-        headers=headers,
-        json={
-            "inputs": prompt,
-        },
-    )
+    retry_count = 0
+    while retry_count < 10:
+        response = requests.post(
+            API_URL,
+            headers=headers,
+            json={
+                "inputs": prompt,
+            },
+        )
 
-    image = Image.open(io.BytesIO(response.content))
-    logger.info(f"Image Generated for prompt:{prompt}")
+        if response.ok:
+            try:
+                image = Image.open(io.BytesIO(response.content))
+                logger.info(f"Image Generated for prompt:{prompt}")
+                image.save(filename)
+                return f"Saved to disk:{filename}"
+            except Exception as e:
+                logger.error(e)
+                break
+        else:
+            try:
+                error = json.loads(response.text)
+                if "estimated_time" in error:
+                    delay = error["estimated_time"]
+                    logger.debug(response.text)
+                    logger.info("Retrying in", delay)
+                    time.sleep(delay)
+                else:
+                    break
+            except Exception as e:
+                logger.error(e)
+                break
 
-    image.save(filename)
+        retry_count += 1
 
-    return f"Saved to disk:{filename}"
+    return f"Error creating image."
 
 
-def generate_image_with_dalle(prompt: str, filename: str, size: int) -> str:
+def generate_image_with_dalle(
+    prompt: str, filename: str, size: int, agent: Agent
+) -> str:
     """Generate an image with DALL-E.
 
     Args:
@@ -102,7 +130,7 @@ def generate_image_with_dalle(prompt: str, filename: str, size: int) -> str:
         n=1,
         size=f"{size}x{size}",
         response_format="b64_json",
-        api_key=CFG.openai_api_key,
+        api_key=agent.config.openai_api_key,
     )
 
     logger.info(f"Image Generated for prompt:{prompt}")
@@ -118,6 +146,7 @@ def generate_image_with_dalle(prompt: str, filename: str, size: int) -> str:
 def generate_image_with_sd_webui(
     prompt: str,
     filename: str,
+    agent: Agent,
     size: int = 512,
     negative_prompt: str = "",
     extra: dict = {},
@@ -134,13 +163,13 @@ def generate_image_with_sd_webui(
     """
     # Create a session and set the basic auth if needed
     s = requests.Session()
-    if CFG.sd_webui_auth:
-        username, password = CFG.sd_webui_auth.split(":")
+    if agent.config.sd_webui_auth:
+        username, password = agent.config.sd_webui_auth.split(":")
         s.auth = (username, password or "")
 
     # Generate the images
     response = requests.post(
-        f"{CFG.sd_webui_url}/sdapi/v1/txt2img",
+        f"{agent.config.sd_webui_url}/sdapi/v1/txt2img",
         json={
             "prompt": prompt,
             "negative_prompt": negative_prompt,

autogpt/commands/improve_code.py

@@ -2,8 +2,9 @@ from __future__ import annotations
 
 import json
 
+from autogpt.agent.agent import Agent
 from autogpt.commands.command import command
-from autogpt.llm import call_ai_function
+from autogpt.llm.utils import call_ai_function
 
 
 @command(
@@ -11,7 +12,7 @@ from autogpt.llm import call_ai_function
     "Get Improved Code",
     '"suggestions": "<list_of_suggestions>", "code": "<full_code_string>"',
 )
-def improve_code(suggestions: list[str], code: str) -> str:
+def improve_code(suggestions: list[str], code: str, agent: Agent) -> str:
     """
     A function that takes in code and suggestions and returns a response from create
       chat completion api call.
@@ -32,4 +33,6 @@ def improve_code(suggestions: list[str], code: str) -> str:
         " provided, making no other changes."
     )
 
-    return call_ai_function(function_string, args, description_string)
+    return call_ai_function(
+        function_string, args, description_string, config=agent.config
+    )

autogpt/commands/task_statuses.py

@@ -0,0 +1,27 @@
+"""Task Statuses module."""
+from __future__ import annotations
+
+from typing import NoReturn
+
+from autogpt.agent.agent import Agent
+from autogpt.commands.command import command
+from autogpt.logs import logger
+
+
+@command(
+    "task_complete",
+    "Task Complete (Shutdown)",
+    '"reason": "<reason>"',
+)
+def task_complete(reason: str, agent: Agent) -> NoReturn:
+    """
+    A function that takes in a string and exits the program
+
+    Parameters:
+        reason (str): The reason for shutting down.
+    Returns:
+        A result string from create chat completion. A list of suggestions to
+            improve the code.
+    """
+    logger.info(title="Shutting down...\n", message=reason)
+    quit()

autogpt/commands/twitter.py

@@ -1,41 +0,0 @@
-"""A module that contains a command to send a tweet."""
-import os
-
-import tweepy
-
-from autogpt.commands.command import command
-
-
-@command(
-    "send_tweet",
-    "Send Tweet",
-    '"tweet_text": "<tweet_text>"',
-)
-def send_tweet(tweet_text: str) -> str:
-    """
-      A function that takes in a string and returns a response from create chat
-        completion api call.
-
-    Args:
-      tweet_text (str): Text to be tweeted.
-
-      Returns:
-          A result from sending the tweet.
-    """
-    consumer_key = os.environ.get("TW_CONSUMER_KEY")
-    consumer_secret = os.environ.get("TW_CONSUMER_SECRET")
-    access_token = os.environ.get("TW_ACCESS_TOKEN")
-    access_token_secret = os.environ.get("TW_ACCESS_TOKEN_SECRET")
-    # Authenticate to Twitter
-    auth = tweepy.OAuthHandler(consumer_key, consumer_secret)
-    auth.set_access_token(access_token, access_token_secret)
-
-    # Create API object
-    api = tweepy.API(auth)
-
-    # Send tweet
-    try:
-        api.update_status(tweet_text)
-        return "Tweet sent successfully!"
-    except tweepy.TweepyException as e:
-        return f"Error sending tweet: {e.reason}"

autogpt/commands/web_requests.py

@@ -1,23 +1,24 @@
 """Browse a webpage and summarize it using the LLM model"""
 from __future__ import annotations
 
+from typing import TYPE_CHECKING
+
 import requests
 from bs4 import BeautifulSoup
 from requests import Response
 
-from autogpt.config import Config
 from autogpt.processing.html import extract_hyperlinks, format_hyperlinks
 from autogpt.url_utils.validators import validate_url
 
-CFG = Config()
-
 session = requests.Session()
-session.headers.update({"User-Agent": CFG.user_agent})
+
+if TYPE_CHECKING:
+    from autogpt.agent.agent import Agent
 
 
 @validate_url
 def get_response(
-    url: str, timeout: int = 10
+    url: str, agent: Agent, timeout: int = 10
 ) -> tuple[None, str] | tuple[Response, None]:
     """Get the response from a URL
 
@@ -33,6 +34,7 @@ def get_response(
         requests.exceptions.RequestException: If the HTTP request fails
     """
     try:
+        session.headers.update({"User-Agent": agent.config.user_agent})
         response = session.get(url, timeout=timeout)
 
         # Check if the response contains an HTTP error
@@ -50,7 +52,7 @@ def get_response(
         return None, f"Error: {str(re)}"
 
 
-def scrape_text(url: str) -> str:
+def scrape_text(url: str, agent: Agent) -> str:
     """Scrape text from a webpage
 
     Args:
@@ -59,7 +61,7 @@ def scrape_text(url: str) -> str:
     Returns:
         str: The scraped text
     """
-    response, error_message = get_response(url)
+    response, error_message = get_response(url, agent)
     if error_message:
         return error_message
     if not response:
@@ -78,7 +80,7 @@ def scrape_text(url: str) -> str:
     return text
 
 
-def scrape_links(url: str) -> str | list[str]:
+def scrape_links(url: str, agent: Agent) -> str | list[str]:
     """Scrape links from a webpage
 
     Args:
@@ -87,7 +89,7 @@ def scrape_links(url: str) -> str | list[str]:
     Returns:
        str | list[str]: The scraped links
     """
-    response, error_message = get_response(url)
+    response, error_message = get_response(url, agent)
     if error_message:
         return error_message
     if not response:
@@ -100,13 +102,3 @@ def scrape_links(url: str) -> str | list[str]:
     hyperlinks = extract_hyperlinks(soup, url)
 
     return format_hyperlinks(hyperlinks)
-
-
-def create_message(chunk, question):
-    """Create a message for the user to summarize a chunk of text"""
-    return {
-        "role": "user",
-        "content": f'"""{chunk}""" Using the above text, answer the following'
-        f' question: "{question}" -- if the question cannot be answered using the'
-        " text, summarize the text.",
-    }

autogpt/commands/web_selenium.py

@@ -4,28 +4,39 @@ from __future__ import annotations
 import logging
 from pathlib import Path
 from sys import platform
+from typing import Optional, Type
 
 from bs4 import BeautifulSoup
-from selenium import webdriver
 from selenium.common.exceptions import WebDriverException
 from selenium.webdriver.chrome.options import Options as ChromeOptions
+from selenium.webdriver.chrome.service import Service as ChromeDriverService
+from selenium.webdriver.chrome.webdriver import WebDriver as ChromeDriver
 from selenium.webdriver.common.by import By
+from selenium.webdriver.edge.options import Options as EdgeOptions
+from selenium.webdriver.edge.service import Service as EdgeDriverService
+from selenium.webdriver.edge.webdriver import WebDriver as EdgeDriver
 from selenium.webdriver.firefox.options import Options as FirefoxOptions
+from selenium.webdriver.firefox.service import Service as GeckoDriverService
+from selenium.webdriver.firefox.webdriver import WebDriver as FirefoxDriver
 from selenium.webdriver.remote.webdriver import WebDriver
 from selenium.webdriver.safari.options import Options as SafariOptions
+from selenium.webdriver.safari.webdriver import WebDriver as SafariDriver
 from selenium.webdriver.support import expected_conditions as EC
 from selenium.webdriver.support.wait import WebDriverWait
 from webdriver_manager.chrome import ChromeDriverManager
 from webdriver_manager.firefox import GeckoDriverManager
+from webdriver_manager.microsoft import EdgeChromiumDriverManager as EdgeDriverManager
 
-import autogpt.processing.text as summary
+from autogpt.agent.agent import Agent
 from autogpt.commands.command import command
-from autogpt.config import Config
+from autogpt.logs import logger
+from autogpt.memory.vector import MemoryItem, get_memory
 from autogpt.processing.html import extract_hyperlinks, format_hyperlinks
 from autogpt.url_utils.validators import validate_url
 
+BrowserOptions = ChromeOptions | EdgeOptions | FirefoxOptions | SafariOptions
+
 FILE_DIR = Path(__file__).parent.parent
-CFG = Config()
 
 
 @command(
@@ -34,7 +45,7 @@ CFG = Config()
     '"url": "<url>", "question": "<what_you_want_to_find_on_website>"',
 )
 @validate_url
-def browse_website(url: str, question: str) -> tuple[str, WebDriver]:
+def browse_website(url: str, question: str, agent: Agent) -> str:
     """Browse a website and return the answer and links to the user
 
     Args:
@@ -45,25 +56,25 @@ def browse_website(url: str, question: str) -> tuple[str, WebDriver]:
         Tuple[str, WebDriver]: The answer and links to the user and the webdriver
     """
     try:
-        driver, text = scrape_text_with_selenium(url)
+        driver, text = scrape_text_with_selenium(url, agent)
     except WebDriverException as e:
         # These errors are often quite long and include lots of context.
         # Just grab the first line.
         msg = e.msg.split("\n")[0]
-        return f"Error: {msg}", None
+        return f"Error: {msg}"
 
     add_header(driver)
-    summary_text = summary.summarize_text(url, text, question, driver)
+    summary = summarize_memorize_webpage(url, text, question, agent, driver)
     links = scrape_links_with_selenium(driver, url)
 
     # Limit links to 5
     if len(links) > 5:
         links = links[:5]
     close_browser(driver)
-    return f"Answer gathered from website: {summary_text} \n \n Links: {links}", driver
+    return f"Answer gathered from website: {summary}\n\nLinks: {links}"
 
 
-def scrape_text_with_selenium(url: str) -> tuple[WebDriver, str]:
+def scrape_text_with_selenium(url: str, agent: Agent) -> tuple[WebDriver, str]:
     """Scrape text from a website using selenium
 
     Args:
@@ -74,44 +85,49 @@ def scrape_text_with_selenium(url: str) -> tuple[WebDriver, str]:
     """
     logging.getLogger("selenium").setLevel(logging.CRITICAL)
 
-    options_available = {
+    options_available: dict[str, Type[BrowserOptions]] = {
         "chrome": ChromeOptions,
-        "safari": SafariOptions,
+        "edge": EdgeOptions,
         "firefox": FirefoxOptions,
+        "safari": SafariOptions,
     }
 
-    options = options_available[CFG.selenium_web_browser]()
+    options: BrowserOptions = options_available[agent.config.selenium_web_browser]()
     options.add_argument(
         "user-agent=Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/112.0.5615.49 Safari/537.36"
     )
 
-    if CFG.selenium_web_browser == "firefox":
-        if CFG.selenium_headless:
+    if agent.config.selenium_web_browser == "firefox":
+        if agent.config.selenium_headless:
             options.headless = True
             options.add_argument("--disable-gpu")
-        driver = webdriver.Firefox(
-            executable_path=GeckoDriverManager().install(), options=options
+        driver = FirefoxDriver(
+            service=GeckoDriverService(GeckoDriverManager().install()), options=options
         )
-    elif CFG.selenium_web_browser == "safari":
+    elif agent.config.selenium_web_browser == "edge":
+        driver = EdgeDriver(
+            service=EdgeDriverService(EdgeDriverManager().install()), options=options
+        )
+    elif agent.config.selenium_web_browser == "safari":
         # Requires a bit more setup on the users end
         # See https://developer.apple.com/documentation/webkit/testing_with_webdriver_in_safari
-        driver = webdriver.Safari(options=options)
+        driver = SafariDriver(options=options)
     else:
         if platform == "linux" or platform == "linux2":
             options.add_argument("--disable-dev-shm-usage")
             options.add_argument("--remote-debugging-port=9222")
 
         options.add_argument("--no-sandbox")
-        if CFG.selenium_headless:
+        if agent.config.selenium_headless:
             options.add_argument("--headless=new")
             options.add_argument("--disable-gpu")
 
         chromium_driver_path = Path("/usr/bin/chromedriver")
 
-        driver = webdriver.Chrome(
-            executable_path=chromium_driver_path
+        driver = ChromeDriver(
+            service=ChromeDriverService(str(chromium_driver_path))
             if chromium_driver_path.exists()
-            else ChromeDriverManager().install(),
+            else ChromeDriverService(ChromeDriverManager().install()),
             options=options,
         )
     driver.get(url)
@@ -175,4 +191,40 @@ def add_header(driver: WebDriver) -> None:
     Returns:
         None
     """
-    driver.execute_script(open(f"{FILE_DIR}/js/overlay.js", "r").read())
+    try:
+        with open(f"{FILE_DIR}/js/overlay.js", "r") as overlay_file:
+            overlay_script = overlay_file.read()
+        driver.execute_script(overlay_script)
+    except Exception as e:
+        print(f"Error executing overlay.js: {e}")
+
+
+def summarize_memorize_webpage(
+    url: str,
+    text: str,
+    question: str,
+    agent: Agent,
+    driver: Optional[WebDriver] = None,
+) -> str:
+    """Summarize text using the OpenAI API
+
+    Args:
+        url (str): The url of the text
+        text (str): The text to summarize
+        question (str): The question to ask the model
+        driver (WebDriver): The webdriver to use to scroll the page
+
+    Returns:
+        str: The summary of the text
+    """
+    if not text:
+        return "Error: No text to summarize"
+
+    text_length = len(text)
+    logger.info(f"Text length: {text_length} characters")
+
+    memory = get_memory(agent.config)
+
+    new_memory = MemoryItem.from_webpage(text, url, question=question)
+    memory.add(new_memory)
+    return new_memory.summary

autogpt/commands/write_tests.py

@@ -3,8 +3,9 @@ from __future__ import annotations
 
 import json
 
+from autogpt.agent.agent import Agent
 from autogpt.commands.command import command
-from autogpt.llm import call_ai_function
+from autogpt.llm.utils import call_ai_function
 
 
 @command(
@@ -12,7 +13,7 @@ from autogpt.llm import call_ai_function
     "Write Tests",
     '"code": "<full_code_string>", "focus": "<list_of_focus_areas>"',
 )
-def write_tests(code: str, focus: list[str]) -> str:
+def write_tests(code: str, focus: list[str], agent: Agent) -> str:
     """
     A function that takes in code and focus topics and returns a response from create
       chat completion api call.
@@ -34,4 +35,6 @@ def write_tests(code: str, focus: list[str]) -> str:
         " specific areas if required."
     )
 
-    return call_ai_function(function_string, args, description_string)
+    return call_ai_function(
+        function_string, args, description_string, config=agent.config
+    )

autogpt/config/ai_config.py

@@ -7,12 +7,14 @@ from __future__ import annotations
 import os
 import platform
 from pathlib import Path
-from typing import Any, Optional, Type
+from typing import TYPE_CHECKING, Optional
 
 import distro
 import yaml
 
-from autogpt.prompts.generator import PromptGenerator
+if TYPE_CHECKING:
+    from autogpt.commands.command import CommandRegistry
+    from autogpt.prompts.generator import PromptGenerator
 
 # Soon this will go in a folder where it remembers more stuff about the run(s)
 SAVE_FILE = str(Path(os.getcwd()) / "ai_settings.yaml")
@@ -53,8 +55,8 @@ class AIConfig:
         self.ai_role = ai_role
         self.ai_goals = ai_goals
         self.api_budget = api_budget
-        self.prompt_generator = None
-        self.command_registry = None
+        self.prompt_generator: PromptGenerator | None = None
+        self.command_registry: CommandRegistry | None = None
 
     @staticmethod
     def load(config_file: str = SAVE_FILE) -> "AIConfig":
@@ -73,7 +75,7 @@ class AIConfig:
 
         try:
             with open(config_file, encoding="utf-8") as file:
-                config_params = yaml.load(file, Loader=yaml.FullLoader)
+                config_params = yaml.load(file, Loader=yaml.FullLoader) or {}
         except FileNotFoundError:
             config_params = {}
 

autogpt/config/config.py

@@ -7,6 +7,7 @@ import yaml
 from auto_gpt_plugin_template import AutoGPTPluginTemplate
 from colorama import Fore
 
+import autogpt
 from autogpt.singleton import Singleton
 
 
@@ -17,8 +18,8 @@ class Config(metaclass=Singleton):
 
     def __init__(self) -> None:
         """Initialize the Config class"""
-        self.workspace_path = None
-        self.file_logger_path = None
+        self.workspace_path: str = None
+        self.file_logger_path: str = None
 
         self.debug_mode = False
         self.continuous_mode = False
@@ -30,17 +31,44 @@ class Config(metaclass=Singleton):
 
         self.authorise_key = os.getenv("AUTHORISE_COMMAND_KEY", "y")
         self.exit_key = os.getenv("EXIT_KEY", "n")
+        self.plain_output = os.getenv("PLAIN_OUTPUT", "False") == "True"
+
+        disabled_command_categories = os.getenv("DISABLED_COMMAND_CATEGORIES")
+        if disabled_command_categories:
+            self.disabled_command_categories = disabled_command_categories.split(",")
+        else:
+            self.disabled_command_categories = []
+
+        self.shell_command_control = os.getenv("SHELL_COMMAND_CONTROL", "denylist")
+
+        # DENY_COMMANDS is deprecated and included for backwards-compatibility
+        shell_denylist = os.getenv("SHELL_DENYLIST", os.getenv("DENY_COMMANDS"))
+        if shell_denylist:
+            self.shell_denylist = shell_denylist.split(",")
+        else:
+            self.shell_denylist = ["sudo", "su"]
+
+        # ALLOW_COMMANDS is deprecated and included for backwards-compatibility
+        shell_allowlist = os.getenv("SHELL_ALLOWLIST", os.getenv("ALLOW_COMMANDS"))
+        if shell_allowlist:
+            self.shell_allowlist = shell_allowlist.split(",")
+        else:
+            self.shell_allowlist = []
+
         self.ai_settings_file = os.getenv("AI_SETTINGS_FILE", "ai_settings.yaml")
+        self.prompt_settings_file = os.getenv(
+            "PROMPT_SETTINGS_FILE", "prompt_settings.yaml"
+        )
         self.fast_llm_model = os.getenv("FAST_LLM_MODEL", "gpt-3.5-turbo")
-        self.smart_llm_model = os.getenv("SMART_LLM_MODEL", "gpt-4")
-        self.fast_token_limit = int(os.getenv("FAST_TOKEN_LIMIT", 4000))
-        self.smart_token_limit = int(os.getenv("SMART_TOKEN_LIMIT", 8000))
-        self.browse_chunk_max_length = int(os.getenv("BROWSE_CHUNK_MAX_LENGTH", 3000))
+        self.smart_llm_model = os.getenv("SMART_LLM_MODEL", "gpt-3.5-turbo")
+        self.embedding_model = os.getenv("EMBEDDING_MODEL", "text-embedding-ada-002")
+
         self.browse_spacy_language_model = os.getenv(
             "BROWSE_SPACY_LANGUAGE_MODEL", "en_core_web_sm"
         )
 
         self.openai_api_key = os.getenv("OPENAI_API_KEY")
+        self.openai_organization = os.getenv("OPENAI_ORGANIZATION")
         self.temperature = float(os.getenv("TEMPERATURE", "0"))
         self.use_azure = os.getenv("USE_AZURE") == "True"
         self.execute_local_commands = (
@@ -55,53 +83,49 @@ class Config(metaclass=Singleton):
             openai.api_type = self.openai_api_type
             openai.api_base = self.openai_api_base
             openai.api_version = self.openai_api_version
+        elif os.getenv("OPENAI_API_BASE_URL", None):
+            openai.api_base = os.getenv("OPENAI_API_BASE_URL")
+
+        if self.openai_organization is not None:
+            openai.organization = self.openai_organization
 
         self.elevenlabs_api_key = os.getenv("ELEVENLABS_API_KEY")
-        self.elevenlabs_voice_1_id = os.getenv("ELEVENLABS_VOICE_1_ID")
-        self.elevenlabs_voice_2_id = os.getenv("ELEVENLABS_VOICE_2_ID")
+        # ELEVENLABS_VOICE_1_ID is deprecated and included for backwards-compatibility
+        self.elevenlabs_voice_id = os.getenv(
+            "ELEVENLABS_VOICE_ID", os.getenv("ELEVENLABS_VOICE_1_ID")
+        )
+        self.streamelements_voice = os.getenv("STREAMELEMENTS_VOICE", "Brian")
 
-        self.use_mac_os_tts = False
-        self.use_mac_os_tts = os.getenv("USE_MAC_OS_TTS")
+        # Backwards-compatibility shim for deprecated env variables
+        if os.getenv("USE_MAC_OS_TTS"):
+            default_tts_provider = "macos"
+        elif self.elevenlabs_api_key:
+            default_tts_provider = "elevenlabs"
+        elif os.getenv("USE_BRIAN_TTS"):
+            default_tts_provider = "streamelements"
+        else:
+            default_tts_provider = "gtts"
 
-        self.chat_messages_enabled = os.getenv("CHAT_MESSAGES_ENABLED") == "True"
-
-        self.use_brian_tts = False
-        self.use_brian_tts = os.getenv("USE_BRIAN_TTS")
+        self.text_to_speech_provider = os.getenv(
+            "TEXT_TO_SPEECH_PROVIDER", default_tts_provider
+        )
 
         self.github_api_key = os.getenv("GITHUB_API_KEY")
         self.github_username = os.getenv("GITHUB_USERNAME")
 
         self.google_api_key = os.getenv("GOOGLE_API_KEY")
-        self.custom_search_engine_id = os.getenv("CUSTOM_SEARCH_ENGINE_ID")
-
-        self.pinecone_api_key = os.getenv("PINECONE_API_KEY")
-        self.pinecone_region = os.getenv("PINECONE_ENV")
-
-        self.weaviate_host = os.getenv("WEAVIATE_HOST")
-        self.weaviate_port = os.getenv("WEAVIATE_PORT")
-        self.weaviate_protocol = os.getenv("WEAVIATE_PROTOCOL", "http")
-        self.weaviate_username = os.getenv("WEAVIATE_USERNAME", None)
-        self.weaviate_password = os.getenv("WEAVIATE_PASSWORD", None)
-        self.weaviate_scopes = os.getenv("WEAVIATE_SCOPES", None)
-        self.weaviate_embedded_path = os.getenv("WEAVIATE_EMBEDDED_PATH")
-        self.weaviate_api_key = os.getenv("WEAVIATE_API_KEY", None)
-        self.use_weaviate_embedded = (
-            os.getenv("USE_WEAVIATE_EMBEDDED", "False") == "True"
+        # CUSTOM_SEARCH_ENGINE_ID is deprecated and included for backwards-compatibility
+        self.google_custom_search_engine_id = os.getenv(
+            "GOOGLE_CUSTOM_SEARCH_ENGINE_ID", os.getenv("CUSTOM_SEARCH_ENGINE_ID")
         )
 
-        # milvus or zilliz cloud configuration.
-        self.milvus_addr = os.getenv("MILVUS_ADDR", "localhost:19530")
-        self.milvus_username = os.getenv("MILVUS_USERNAME")
-        self.milvus_password = os.getenv("MILVUS_PASSWORD")
-        self.milvus_collection = os.getenv("MILVUS_COLLECTION", "autogpt")
-        self.milvus_secure = os.getenv("MILVUS_SECURE") == "True"
-
         self.image_provider = os.getenv("IMAGE_PROVIDER")
         self.image_size = int(os.getenv("IMAGE_SIZE", 256))
         self.huggingface_api_token = os.getenv("HUGGINGFACE_API_TOKEN")
         self.huggingface_image_model = os.getenv(
             "HUGGINGFACE_IMAGE_MODEL", "CompVis/stable-diffusion-v1-4"
         )
+        self.audio_to_text_provider = os.getenv("AUDIO_TO_TEXT_PROVIDER", "huggingface")
         self.huggingface_audio_to_text_model = os.getenv(
             "HUGGINGFACE_AUDIO_TO_TEXT_MODEL"
         )
@@ -121,25 +145,48 @@ class Config(metaclass=Singleton):
             " (KHTML, like Gecko) Chrome/83.0.4103.97 Safari/537.36",
         )
 
+        self.memory_backend = os.getenv("MEMORY_BACKEND", "json_file")
+        self.memory_index = os.getenv("MEMORY_INDEX", "auto-gpt-memory")
+
         self.redis_host = os.getenv("REDIS_HOST", "localhost")
-        self.redis_port = os.getenv("REDIS_PORT", "6379")
+        self.redis_port = int(os.getenv("REDIS_PORT", "6379"))
         self.redis_password = os.getenv("REDIS_PASSWORD", "")
         self.wipe_redis_on_start = os.getenv("WIPE_REDIS_ON_START", "True") == "True"
-        self.memory_index = os.getenv("MEMORY_INDEX", "auto-gpt")
-        # Note that indexes must be created on db 0 in redis, this is not configurable.
-
-        self.memory_backend = os.getenv("MEMORY_BACKEND", "local")
 
         self.plugins_dir = os.getenv("PLUGINS_DIR", "plugins")
         self.plugins: List[AutoGPTPluginTemplate] = []
         self.plugins_openai = []
 
+        # Deprecated. Kept for backwards-compatibility. Will remove in a future version.
         plugins_allowlist = os.getenv("ALLOWLISTED_PLUGINS")
         if plugins_allowlist:
             self.plugins_allowlist = plugins_allowlist.split(",")
         else:
             self.plugins_allowlist = []
-        self.plugins_denylist = []
+
+        # Deprecated. Kept for backwards-compatibility. Will remove in a future version.
+        plugins_denylist = os.getenv("DENYLISTED_PLUGINS")
+        if plugins_denylist:
+            self.plugins_denylist = plugins_denylist.split(",")
+        else:
+            self.plugins_denylist = []
+
+        # Avoid circular imports
+        from autogpt.plugins import DEFAULT_PLUGINS_CONFIG_FILE
+
+        self.plugins_config_file = os.getenv(
+            "PLUGINS_CONFIG_FILE", DEFAULT_PLUGINS_CONFIG_FILE
+        )
+        self.load_plugins_config()
+
+        self.chat_messages_enabled = os.getenv("CHAT_MESSAGES_ENABLED") == "True"
+
+    def load_plugins_config(self) -> "autogpt.plugins.PluginsConfig":
+        # Avoid circular import
+        from autogpt.plugins.plugins_config import PluginsConfig
+
+        self.plugins_config = PluginsConfig.load_config(global_config=self)
+        return self.plugins_config
 
     def get_azure_deployment_id_for_model(self, model: str) -> str:
         """
@@ -180,7 +227,7 @@ class Config(metaclass=Singleton):
             None
         """
         with open(config_file) as file:
-            config_params = yaml.load(file, Loader=yaml.FullLoader)
+            config_params = yaml.load(file, Loader=yaml.FullLoader) or {}
         self.openai_api_type = config_params.get("azure_api_type") or "azure"
         self.openai_api_base = config_params.get("azure_api_base") or ""
         self.openai_api_version = (
@@ -208,17 +255,9 @@ class Config(metaclass=Singleton):
         """Set the smart LLM model value."""
         self.smart_llm_model = value
 
-    def set_fast_token_limit(self, value: int) -> None:
-        """Set the fast token limit value."""
-        self.fast_token_limit = value
-
-    def set_smart_token_limit(self, value: int) -> None:
-        """Set the smart token limit value."""
-        self.smart_token_limit = value
-
-    def set_browse_chunk_max_length(self, value: int) -> None:
-        """Set the browse_website command chunk max length value."""
-        self.browse_chunk_max_length = value
+    def set_embedding_model(self, value: str) -> None:
+        """Set the model to use for creating embeddings."""
+        self.embedding_model = value
 
     def set_openai_api_key(self, value: str) -> None:
         """Set the OpenAI API key value."""
@@ -230,7 +269,7 @@ class Config(metaclass=Singleton):
 
     def set_elevenlabs_voice_1_id(self, value: str) -> None:
         """Set the ElevenLabs Voice 1 ID value."""
-        self.elevenlabs_voice_1_id = value
+        self.elevenlabs_voice_id = value
 
     def set_elevenlabs_voice_2_id(self, value: str) -> None:
         """Set the ElevenLabs Voice 2 ID value."""
@@ -242,15 +281,7 @@ class Config(metaclass=Singleton):
 
     def set_custom_search_engine_id(self, value: str) -> None:
         """Set the custom search engine id value."""
-        self.custom_search_engine_id = value
-
-    def set_pinecone_api_key(self, value: str) -> None:
-        """Set the Pinecone API key value."""
-        self.pinecone_api_key = value
-
-    def set_pinecone_region(self, value: str) -> None:
-        """Set the Pinecone region value."""
-        self.pinecone_region = value
+        self.google_custom_search_engine_id = value
 
     def set_debug_mode(self, value: bool) -> None:
         """Set the debug mode value."""

autogpt/config/prompt_config.py

@@ -0,0 +1,53 @@
+# sourcery skip: do-not-use-staticmethod
+"""
+A module that contains the PromptConfig class object that contains the configuration
+"""
+import yaml
+from colorama import Fore
+
+from autogpt import utils
+from autogpt.config.config import Config
+from autogpt.logs import logger
+
+CFG = Config()
+
+
+class PromptConfig:
+    """
+    A class object that contains the configuration information for the prompt, which will be used by the prompt generator
+
+    Attributes:
+        constraints (list): Constraints list for the prompt generator.
+        resources (list): Resources list for the prompt generator.
+        performance_evaluations (list): Performance evaluation list for the prompt generator.
+    """
+
+    def __init__(
+        self,
+        config_file: str = CFG.prompt_settings_file,
+    ) -> None:
+        """
+        Initialize a class instance with parameters (constraints, resources, performance_evaluations) loaded from
+          yaml file if yaml file exists,
+        else raises error.
+
+        Parameters:
+            constraints (list): Constraints list for the prompt generator.
+            resources (list): Resources list for the prompt generator.
+            performance_evaluations (list): Performance evaluation list for the prompt generator.
+        Returns:
+            None
+        """
+        # Validate file
+        (validated, message) = utils.validate_yaml_file(config_file)
+        if not validated:
+            logger.typewriter_log("FAILED FILE VALIDATION", Fore.RED, message)
+            logger.double_check()
+            exit(1)
+
+        with open(config_file, encoding="utf-8") as file:
+            config_params = yaml.load(file, Loader=yaml.FullLoader)
+
+        self.constraints = config_params.get("constraints", [])
+        self.resources = config_params.get("resources", [])
+        self.performance_evaluations = config_params.get("performance_evaluations", [])

autogpt/configurator.py

@@ -1,19 +1,29 @@
 """Configurator module."""
+from __future__ import annotations
+
+from typing import TYPE_CHECKING
+
 import click
 from colorama import Back, Fore, Style
 
 from autogpt import utils
-from autogpt.config import Config
+from autogpt.llm.utils import check_model
 from autogpt.logs import logger
-from autogpt.memory import get_supported_memory_backends
+from autogpt.memory.vector import get_supported_memory_backends
 
-CFG = Config()
+if TYPE_CHECKING:
+    from autogpt.config import Config
+
+GPT_4_MODEL = "gpt-4"
+GPT_3_MODEL = "gpt-3.5-turbo"
 
 
 def create_config(
+    config: Config,
     continuous: bool,
     continuous_limit: int,
     ai_settings_file: str,
+    prompt_settings_file: str,
     skip_reprompt: bool,
     speak: bool,
     debug: bool,
@@ -30,6 +40,7 @@ def create_config(
         continuous (bool): Whether to run in continuous mode
         continuous_limit (int): The number of times to run in continuous mode
         ai_settings_file (str): The path to the ai_settings.yaml file
+        prompt_settings_file (str): The path to the prompt_settings.yaml file
         skip_reprompt (bool): Whether to skip the re-prompting messages at the beginning of the script
         speak (bool): Whether to enable speak mode
         debug (bool): Whether to enable debug mode
@@ -40,13 +51,13 @@ def create_config(
         allow_downloads (bool): Whether to allow Auto-GPT to download files natively
         skips_news (bool): Whether to suppress the output of latest news on startup
     """
-    CFG.set_debug_mode(False)
-    CFG.set_continuous_mode(False)
-    CFG.set_speak_mode(False)
+    config.set_debug_mode(False)
+    config.set_continuous_mode(False)
+    config.set_speak_mode(False)
 
     if debug:
         logger.typewriter_log("Debug Mode: ", Fore.GREEN, "ENABLED")
-        CFG.set_debug_mode(True)
+        config.set_debug_mode(True)
 
     if continuous:
         logger.typewriter_log("Continuous Mode: ", Fore.RED, "ENABLED")
@@ -57,13 +68,13 @@ def create_config(
             " cause your AI to run forever or carry out actions you would not usually"
             " authorise. Use at your own risk.",
         )
-        CFG.set_continuous_mode(True)
+        config.set_continuous_mode(True)
 
         if continuous_limit:
             logger.typewriter_log(
                 "Continuous Limit: ", Fore.GREEN, f"{continuous_limit}"
             )
-            CFG.set_continuous_limit(continuous_limit)
+            config.set_continuous_limit(continuous_limit)
 
     # Check if continuous limit is used without continuous mode
     if continuous_limit and not continuous:
@@ -71,15 +82,28 @@ def create_config(
 
     if speak:
         logger.typewriter_log("Speak Mode: ", Fore.GREEN, "ENABLED")
-        CFG.set_speak_mode(True)
+        config.set_speak_mode(True)
 
+    # Set the default LLM models
     if gpt3only:
         logger.typewriter_log("GPT3.5 Only Mode: ", Fore.GREEN, "ENABLED")
-        CFG.set_smart_llm_model(CFG.fast_llm_model)
+        # --gpt3only should always use gpt-3.5-turbo, despite user's FAST_LLM_MODEL config
+        config.set_fast_llm_model(GPT_3_MODEL)
+        config.set_smart_llm_model(GPT_3_MODEL)
 
-    if gpt4only:
+    elif (
+        gpt4only
+        and check_model(GPT_4_MODEL, model_type="smart_llm_model") == GPT_4_MODEL
+    ):
         logger.typewriter_log("GPT4 Only Mode: ", Fore.GREEN, "ENABLED")
-        CFG.set_fast_llm_model(CFG.smart_llm_model)
+        # --gpt4only should always use gpt-4, despite user's SMART_LLM_MODEL config
+        config.set_fast_llm_model(GPT_4_MODEL)
+        config.set_smart_llm_model(GPT_4_MODEL)
+    else:
+        config.set_fast_llm_model(check_model(config.fast_llm_model, "fast_llm_model"))
+        config.set_smart_llm_model(
+            check_model(config.smart_llm_model, "smart_llm_model")
+        )
 
     if memory_type:
         supported_memory = get_supported_memory_backends()
@@ -90,13 +114,13 @@ def create_config(
                 Fore.RED,
                 f"{supported_memory}",
             )
-            logger.typewriter_log("Defaulting to: ", Fore.YELLOW, CFG.memory_backend)
+            logger.typewriter_log("Defaulting to: ", Fore.YELLOW, config.memory_backend)
         else:
-            CFG.memory_backend = chosen
+            config.memory_backend = chosen
 
     if skip_reprompt:
         logger.typewriter_log("Skip Re-prompt: ", Fore.GREEN, "ENABLED")
-        CFG.skip_reprompt = True
+        config.skip_reprompt = True
 
     if ai_settings_file:
         file = ai_settings_file
@@ -109,11 +133,24 @@ def create_config(
             exit(1)
 
         logger.typewriter_log("Using AI Settings File:", Fore.GREEN, file)
-        CFG.ai_settings_file = file
-        CFG.skip_reprompt = True
+        config.ai_settings_file = file
+        config.skip_reprompt = True
+
+    if prompt_settings_file:
+        file = prompt_settings_file
+
+        # Validate file
+        (validated, message) = utils.validate_yaml_file(file)
+        if not validated:
+            logger.typewriter_log("FAILED FILE VALIDATION", Fore.RED, message)
+            logger.double_check()
+            exit(1)
+
+        logger.typewriter_log("Using Prompt Settings File:", Fore.GREEN, file)
+        config.prompt_settings_file = file
 
     if browser_name:
-        CFG.selenium_web_browser = browser_name
+        config.selenium_web_browser = browser_name
 
     if allow_downloads:
         logger.typewriter_log("Native Downloading:", Fore.GREEN, "ENABLED")
@@ -128,7 +165,7 @@ def create_config(
             Fore.YELLOW,
             f"{Back.RED + Style.BRIGHT}ALWAYS REMEMBER TO NEVER OPEN FILES YOU AREN'T SURE OF!{Style.RESET_ALL}",
         )
-        CFG.allow_downloads = True
+        config.allow_downloads = True
 
     if skip_news:
-        CFG.skip_news = True
+        config.skip_news = True

autogpt/json_utils/json_fix_general.py

@@ -1,121 +0,0 @@
-"""This module contains functions to fix JSON strings using general programmatic approaches, suitable for addressing
-common JSON formatting issues."""
-from __future__ import annotations
-
-import contextlib
-import json
-import re
-from typing import Optional
-
-from autogpt.config import Config
-from autogpt.json_utils.utilities import extract_char_position
-from autogpt.logs import logger
-
-CFG = Config()
-
-
-def fix_invalid_escape(json_to_load: str, error_message: str) -> str:
-    """Fix invalid escape sequences in JSON strings.
-
-    Args:
-        json_to_load (str): The JSON string.
-        error_message (str): The error message from the JSONDecodeError
-          exception.
-
-    Returns:
-        str: The JSON string with invalid escape sequences fixed.
-    """
-    while error_message.startswith("Invalid \\escape"):
-        bad_escape_location = extract_char_position(error_message)
-        json_to_load = (
-            json_to_load[:bad_escape_location] + json_to_load[bad_escape_location + 1 :]
-        )
-        try:
-            json.loads(json_to_load)
-            return json_to_load
-        except json.JSONDecodeError as e:
-            logger.debug("json loads error - fix invalid escape", e)
-            error_message = str(e)
-    return json_to_load
-
-
-def balance_braces(json_string: str) -> Optional[str]:
-    """
-    Balance the braces in a JSON string.
-
-    Args:
-        json_string (str): The JSON string.
-
-    Returns:
-        str: The JSON string with braces balanced.
-    """
-
-    open_braces_count = json_string.count("{")
-    close_braces_count = json_string.count("}")
-
-    while open_braces_count > close_braces_count:
-        json_string += "}"
-        close_braces_count += 1
-
-    while close_braces_count > open_braces_count:
-        json_string = json_string.rstrip("}")
-        close_braces_count -= 1
-
-    with contextlib.suppress(json.JSONDecodeError):
-        json.loads(json_string)
-        return json_string
-
-
-def add_quotes_to_property_names(json_string: str) -> str:
-    """
-    Add quotes to property names in a JSON string.
-
-    Args:
-        json_string (str): The JSON string.
-
-    Returns:
-        str: The JSON string with quotes added to property names.
-    """
-
-    def replace_func(match: re.Match) -> str:
-        return f'"{match[1]}":'
-
-    property_name_pattern = re.compile(r"(\w+):")
-    corrected_json_string = property_name_pattern.sub(replace_func, json_string)
-
-    try:
-        json.loads(corrected_json_string)
-        return corrected_json_string
-    except json.JSONDecodeError as e:
-        raise e
-
-
-def correct_json(json_to_load: str) -> str:
-    """
-    Correct common JSON errors.
-    Args:
-        json_to_load (str): The JSON string.
-    """
-
-    try:
-        logger.debug("json", json_to_load)
-        json.loads(json_to_load)
-        return json_to_load
-    except json.JSONDecodeError as e:
-        logger.debug("json loads error", e)
-        error_message = str(e)
-        if error_message.startswith("Invalid \\escape"):
-            json_to_load = fix_invalid_escape(json_to_load, error_message)
-        if error_message.startswith(
-            "Expecting property name enclosed in double quotes"
-        ):
-            json_to_load = add_quotes_to_property_names(json_to_load)
-            try:
-                json.loads(json_to_load)
-                return json_to_load
-            except json.JSONDecodeError as e:
-                logger.debug("json loads error - add quotes", e)
-                error_message = str(e)
-        if balanced_str := balance_braces(json_to_load):
-            return balanced_str
-    return json_to_load

autogpt/json_utils/json_fix_llm.py

@@ -1,239 +0,0 @@
-"""This module contains functions to fix JSON strings generated by LLM models, such as ChatGPT, using the assistance
-of the ChatGPT API or LLM models."""
-from __future__ import annotations
-
-import contextlib
-import json
-from typing import Any, Dict
-
-from colorama import Fore
-from regex import regex
-
-from autogpt.config import Config
-from autogpt.json_utils.json_fix_general import correct_json
-from autogpt.llm import call_ai_function
-from autogpt.logs import logger
-from autogpt.speech import say_text
-
-JSON_SCHEMA = """
-{
-    "command": {
-        "name": "command name",
-        "args": {
-            "arg name": "value"
-        }
-    },
-    "thoughts":
-    {
-        "text": "thought",
-        "reasoning": "reasoning",
-        "plan": "- short bulleted\n- list that conveys\n- long-term plan",
-        "criticism": "constructive self-criticism",
-        "speak": "thoughts summary to say to user"
-    }
-}
-"""
-
-CFG = Config()
-
-
-def auto_fix_json(json_string: str, schema: str) -> str:
-    """Fix the given JSON string to make it parseable and fully compliant with
-        the provided schema using GPT-3.
-
-    Args:
-        json_string (str): The JSON string to fix.
-        schema (str): The schema to use to fix the JSON.
-    Returns:
-        str: The fixed JSON string.
-    """
-    # Try to fix the JSON using GPT:
-    function_string = "def fix_json(json_string: str, schema:str=None) -> str:"
-    args = [f"'''{json_string}'''", f"'''{schema}'''"]
-    description_string = (
-        "This function takes a JSON string and ensures that it"
-        " is parseable and fully compliant with the provided schema. If an object"
-        " or field specified in the schema isn't contained within the correct JSON,"
-        " it is omitted. The function also escapes any double quotes within JSON"
-        " string values to ensure that they are valid. If the JSON string contains"
-        " any None or NaN values, they are replaced with null before being parsed."
-    )
-
-    # If it doesn't already start with a "`", add one:
-    if not json_string.startswith("`"):
-        json_string = "```json\n" + json_string + "\n```"
-    result_string = call_ai_function(
-        function_string, args, description_string, model=CFG.fast_llm_model
-    )
-    logger.debug("------------ JSON FIX ATTEMPT ---------------")
-    logger.debug(f"Original JSON: {json_string}")
-    logger.debug("-----------")
-    logger.debug(f"Fixed JSON: {result_string}")
-    logger.debug("----------- END OF FIX ATTEMPT ----------------")
-
-    try:
-        json.loads(result_string)  # just check the validity
-        return result_string
-    except json.JSONDecodeError:  # noqa: E722
-        # Get the call stack:
-        # import traceback
-        # call_stack = traceback.format_exc()
-        # print(f"Failed to fix JSON: '{json_string}' "+call_stack)
-        return "failed"
-
-
-def fix_json_using_multiple_techniques(assistant_reply: str) -> Dict[Any, Any]:
-    """Fix the given JSON string to make it parseable and fully compliant with two techniques.
-
-    Args:
-        json_string (str): The JSON string to fix.
-
-    Returns:
-        str: The fixed JSON string.
-    """
-    assistant_reply = assistant_reply.strip()
-    if assistant_reply.startswith("```json"):
-        assistant_reply = assistant_reply[7:]
-    if assistant_reply.endswith("```"):
-        assistant_reply = assistant_reply[:-3]
-    try:
-        return json.loads(assistant_reply)  # just check the validity
-    except json.JSONDecodeError:  # noqa: E722
-        pass
-
-    if assistant_reply.startswith("json "):
-        assistant_reply = assistant_reply[5:]
-        assistant_reply = assistant_reply.strip()
-    try:
-        return json.loads(assistant_reply)  # just check the validity
-    except json.JSONDecodeError:  # noqa: E722
-        pass
-
-    # Parse and print Assistant response
-    assistant_reply_json = fix_and_parse_json(assistant_reply)
-    logger.debug("Assistant reply JSON: %s", str(assistant_reply_json))
-    if assistant_reply_json == {}:
-        assistant_reply_json = attempt_to_fix_json_by_finding_outermost_brackets(
-            assistant_reply
-        )
-
-    logger.debug("Assistant reply JSON 2: %s", str(assistant_reply_json))
-    if assistant_reply_json != {}:
-        return assistant_reply_json
-
-    logger.error(
-        "Error: The following AI output couldn't be converted to a JSON:\n",
-        assistant_reply,
-    )
-    if CFG.speak_mode:
-        say_text("I have received an invalid JSON response from the OpenAI API.")
-
-    return {}
-
-
-def fix_and_parse_json(
-    json_to_load: str, try_to_fix_with_gpt: bool = True
-) -> Dict[Any, Any]:
-    """Fix and parse JSON string
-
-    Args:
-        json_to_load (str): The JSON string.
-        try_to_fix_with_gpt (bool, optional): Try to fix the JSON with GPT.
-            Defaults to True.
-
-    Returns:
-        str or dict[Any, Any]: The parsed JSON.
-    """
-
-    with contextlib.suppress(json.JSONDecodeError):
-        json_to_load = json_to_load.replace("\t", "")
-        return json.loads(json_to_load)
-
-    with contextlib.suppress(json.JSONDecodeError):
-        json_to_load = correct_json(json_to_load)
-        return json.loads(json_to_load)
-    # Let's do something manually:
-    # sometimes GPT responds with something BEFORE the braces:
-    # "I'm sorry, I don't understand. Please try again."
-    # {"text": "I'm sorry, I don't understand. Please try again.",
-    #  "confidence": 0.0}
-    # So let's try to find the first brace and then parse the rest
-    #  of the string
-    try:
-        brace_index = json_to_load.index("{")
-        maybe_fixed_json = json_to_load[brace_index:]
-        last_brace_index = maybe_fixed_json.rindex("}")
-        maybe_fixed_json = maybe_fixed_json[: last_brace_index + 1]
-        return json.loads(maybe_fixed_json)
-    except (json.JSONDecodeError, ValueError) as e:
-        return try_ai_fix(try_to_fix_with_gpt, e, json_to_load)
-
-
-def try_ai_fix(
-    try_to_fix_with_gpt: bool, exception: Exception, json_to_load: str
-) -> Dict[Any, Any]:
-    """Try to fix the JSON with the AI
-
-    Args:
-        try_to_fix_with_gpt (bool): Whether to try to fix the JSON with the AI.
-        exception (Exception): The exception that was raised.
-        json_to_load (str): The JSON string to load.
-
-    Raises:
-        exception: If try_to_fix_with_gpt is False.
-
-    Returns:
-        str or dict[Any, Any]: The JSON string or dictionary.
-    """
-    if not try_to_fix_with_gpt:
-        raise exception
-    if CFG.debug_mode:
-        logger.warn(
-            "Warning: Failed to parse AI output, attempting to fix."
-            "\n If you see this warning frequently, it's likely that"
-            " your prompt is confusing the AI. Try changing it up"
-            " slightly."
-        )
-    # Now try to fix this up using the ai_functions
-    ai_fixed_json = auto_fix_json(json_to_load, JSON_SCHEMA)
-
-    if ai_fixed_json != "failed":
-        return json.loads(ai_fixed_json)
-    # This allows the AI to react to the error message,
-    #   which usually results in it correcting its ways.
-    # logger.error("Failed to fix AI output, telling the AI.")
-    return {}
-
-
-def attempt_to_fix_json_by_finding_outermost_brackets(json_string: str):
-    if CFG.speak_mode and CFG.debug_mode:
-        say_text(
-            "I have received an invalid JSON response from the OpenAI API. "
-            "Trying to fix it now."
-        )
-        logger.error("Attempting to fix JSON by finding outermost brackets\n")
-
-    try:
-        json_pattern = regex.compile(r"\{(?:[^{}]|(?R))*\}")
-        json_match = json_pattern.search(json_string)
-
-        if json_match:
-            # Extract the valid JSON object from the string
-            json_string = json_match.group(0)
-            logger.typewriter_log(
-                title="Apparently json was fixed.", title_color=Fore.GREEN
-            )
-            if CFG.speak_mode and CFG.debug_mode:
-                say_text("Apparently json was fixed.")
-        else:
-            return {}
-
-    except (json.JSONDecodeError, ValueError):
-        if CFG.debug_mode:
-            logger.error(f"Error: Invalid JSON: {json_string}\n")
-        if CFG.speak_mode:
-            say_text("Didn't work. I will have to ignore this response then.")
-        logger.error("Error: Invalid JSON, setting it to empty JSON now.\n")
-        json_string = {}
-
-    return fix_and_parse_json(json_string)

autogpt/json_utils/llm_response_format_1.json

@@ -5,11 +5,25 @@
         "thoughts": {
             "type": "object",
             "properties": {
-                "text": {"type": "string"},
-                "reasoning": {"type": "string"},
-                "plan": {"type": "string"},
-                "criticism": {"type": "string"},
-                "speak": {"type": "string"}
+                "text": {
+                    "type": "string",
+                    "description": "thoughts"
+                },
+                "reasoning": {
+                    "type": "string"
+                },
+                "plan": {
+                    "type": "string",
+                    "description": "- short bulleted\n- list that conveys\n- long-term plan"
+                },
+                "criticism": {
+                    "type": "string",
+                    "description": "constructive self-criticism"
+                },
+                "speak": {
+                    "type": "string",
+                    "description": "thoughts summary to say to user"
+                }
             },
             "required": ["text", "reasoning", "plan", "criticism", "speak"],
             "additionalProperties": false

autogpt/json_utils/utilities.py

@@ -1,6 +1,8 @@
 """Utilities for the json_fixes package."""
+import ast
 import json
-import re
+import os.path
+from typing import Any
 
 from jsonschema import Draft7Validator
 
@@ -11,36 +13,47 @@ CFG = Config()
 LLM_DEFAULT_RESPONSE_FORMAT = "llm_response_format_1"
 
 
-def extract_char_position(error_message: str) -> int:
-    """Extract the character position from the JSONDecodeError message.
+def extract_json_from_response(response_content: str) -> dict:
+    # Sometimes the response includes the JSON in a code block with ```
+    if response_content.startswith("```") and response_content.endswith("```"):
+        # Discard the first and last ```, then re-join in case the response naturally included ```
+        response_content = "```".join(response_content.split("```")[1:-1])
 
-    Args:
-        error_message (str): The error message from the JSONDecodeError
-          exception.
-
-    Returns:
-        int: The character position.
-    """
-
-    char_pattern = re.compile(r"\(char (\d+)\)")
-    if match := char_pattern.search(error_message):
-        return int(match[1])
-    else:
-        raise ValueError("Character position not found in the error message.")
+    # response content comes from OpenAI as a Python `str(content_dict)`, literal_eval reverses this
+    try:
+        return ast.literal_eval(response_content)
+    except BaseException as e:
+        logger.error(f"Error parsing JSON response with literal_eval {e}")
+        # TODO: How to raise an error here without causing the program to exit?
+        return {}
 
 
-def validate_json(json_object: object, schema_name: str) -> dict | None:
+def llm_response_schema(
+    schema_name: str = LLM_DEFAULT_RESPONSE_FORMAT,
+) -> dict[str, Any]:
+    filename = os.path.join(os.path.dirname(__file__), f"{schema_name}.json")
+    with open(filename, "r") as f:
+        return json.load(f)
+
+
+def validate_json(
+    json_object: object, schema_name: str = LLM_DEFAULT_RESPONSE_FORMAT
+) -> bool:
     """
     :type schema_name: object
     :param schema_name: str
     :type json_object: object
+
+    Returns:
+        bool: Whether the json_object is valid or not
     """
-    with open(f"autogpt/json_utils/{schema_name}.json", "r") as f:
-        schema = json.load(f)
+    schema = llm_response_schema(schema_name)
     validator = Draft7Validator(schema)
 
     if errors := sorted(validator.iter_errors(json_object), key=lambda e: e.path):
-        logger.error("The JSON object is invalid.")
+        for error in errors:
+            logger.error(f"JSON Validation Error: {error}")
+
         if CFG.debug_mode:
             logger.error(
                 json.dumps(json_object, indent=4)
@@ -49,10 +62,11 @@ def validate_json(json_object: object, schema_name: str) -> dict | None:
 
             for error in errors:
                 logger.error(f"Error: {error.message}")
-    else:
-        logger.debug("The JSON object is valid.")
+        return False
 
-    return json_object
+    logger.debug("The JSON object is valid.")
+
+    return True
 
 
 def validate_json_string(json_string: str, schema_name: str) -> dict | None:
@@ -64,7 +78,9 @@ def validate_json_string(json_string: str, schema_name: str) -> dict | None:
 
     try:
         json_loaded = json.loads(json_string)
-        return validate_json(json_loaded, schema_name)
+        if not validate_json(json_loaded, schema_name):
+            return None
+        return json_loaded
     except:
         return None
 

autogpt/llm/__init__.py

@@ -1,4 +1,3 @@
-from autogpt.llm.api_manager import ApiManager
 from autogpt.llm.base import (
     ChatModelInfo,
     ChatModelResponse,
@@ -8,17 +7,8 @@ from autogpt.llm.base import (
     Message,
     ModelInfo,
 )
-from autogpt.llm.chat import chat_with_ai, create_chat_message, generate_context
-from autogpt.llm.llm_utils import (
-    call_ai_function,
-    create_chat_completion,
-    get_ada_embedding,
-)
-from autogpt.llm.modelsinfo import COSTS
-from autogpt.llm.token_counter import count_message_tokens, count_string_tokens
 
 __all__ = [
-    "ApiManager",
     "Message",
     "ModelInfo",
     "ChatModelInfo",
@@ -26,13 +16,4 @@ __all__ = [
     "LLMResponse",
     "ChatModelResponse",
     "EmbeddingModelResponse",
-    "create_chat_message",
-    "generate_context",
-    "chat_with_ai",
-    "call_ai_function",
-    "create_chat_completion",
-    "get_ada_embedding",
-    "COSTS",
-    "count_message_tokens",
-    "count_string_tokens",
 ]

autogpt/llm/api_manager.py

@@ -1,9 +1,13 @@
 from __future__ import annotations
 
+from typing import List, Optional
+
 import openai
+from openai import Model
 
 from autogpt.config import Config
-from autogpt.llm.modelsinfo import COSTS
+from autogpt.llm.base import CompletionModelInfo, MessageDict
+from autogpt.llm.providers.openai import OPEN_AI_MODELS
 from autogpt.logs import logger
 from autogpt.singleton import Singleton
 
@@ -14,21 +18,23 @@ class ApiManager(metaclass=Singleton):
         self.total_completion_tokens = 0
         self.total_cost = 0
         self.total_budget = 0
+        self.models: Optional[list[Model]] = None
 
     def reset(self):
         self.total_prompt_tokens = 0
         self.total_completion_tokens = 0
         self.total_cost = 0
         self.total_budget = 0.0
+        self.models = None
 
     def create_chat_completion(
         self,
-        messages: list,  # type: ignore
+        messages: list[MessageDict],
         model: str | None = None,
         temperature: float = None,
         max_tokens: int | None = None,
         deployment_id=None,
-    ) -> str:
+    ):
         """
         Create a chat completion and update the cost.
         Args:
@@ -59,13 +65,14 @@ class ApiManager(metaclass=Singleton):
                 max_tokens=max_tokens,
                 api_key=cfg.openai_api_key,
             )
-        logger.debug(f"Response: {response}")
-        prompt_tokens = response.usage.prompt_tokens
-        completion_tokens = response.usage.completion_tokens
-        self.update_cost(prompt_tokens, completion_tokens, model)
+        if not hasattr(response, "error"):
+            logger.debug(f"Response: {response}")
+            prompt_tokens = response.usage.prompt_tokens
+            completion_tokens = response.usage.completion_tokens
+            self.update_cost(prompt_tokens, completion_tokens, model)
         return response
 
-    def update_cost(self, prompt_tokens, completion_tokens, model):
+    def update_cost(self, prompt_tokens, completion_tokens, model: str):
         """
         Update the total cost, prompt tokens, and completion tokens.
 
@@ -74,12 +81,18 @@ class ApiManager(metaclass=Singleton):
         completion_tokens (int): The number of tokens used in the completion.
         model (str): The model used for the API call.
         """
+        # the .model property in API responses can contain version suffixes like -v2
+        model = model[:-3] if model.endswith("-v2") else model
+        model_info = OPEN_AI_MODELS[model]
+
         self.total_prompt_tokens += prompt_tokens
         self.total_completion_tokens += completion_tokens
-        self.total_cost += (
-            prompt_tokens * COSTS[model]["prompt"]
-            + completion_tokens * COSTS[model]["completion"]
-        ) / 1000
+        self.total_cost += prompt_tokens * model_info.prompt_token_cost / 1000
+        if issubclass(type(model_info), CompletionModelInfo):
+            self.total_cost += (
+                completion_tokens * model_info.completion_token_cost / 1000
+            )
+
         logger.debug(f"Total running cost: ${self.total_cost:.3f}")
 
     def set_total_budget(self, total_budget):
@@ -126,3 +139,17 @@ class ApiManager(metaclass=Singleton):
         float: The total budget for API calls.
         """
         return self.total_budget
+
+    def get_models(self) -> List[Model]:
+        """
+        Get list of available GPT models.
+
+        Returns:
+        list: List of available GPT models.
+
+        """
+        if self.models is None:
+            all_models = openai.Model.list()["data"]
+            self.models = [model for model in all_models if "gpt" in model["id"]]
+
+        return self.models

autogpt/llm/base.py

@@ -1,12 +1,28 @@
+from __future__ import annotations
+
 from dataclasses import dataclass, field
-from typing import List, TypedDict
+from math import ceil, floor
+from typing import List, Literal, TypedDict
+
+MessageRole = Literal["system", "user", "assistant"]
+MessageType = Literal["ai_response", "action_result"]
 
 
-class Message(TypedDict):
+class MessageDict(TypedDict):
+    role: MessageRole
+    content: str
+
+
+@dataclass
+class Message:
     """OpenAI Message object containing a role and the message content"""
 
-    role: str
+    role: MessageRole
     content: str
+    type: MessageType | None = None
+
+    def raw(self) -> MessageDict:
+        return {"role": self.role, "content": self.content}
 
 
 @dataclass
@@ -15,20 +31,28 @@ class ModelInfo:
 
     Would be lovely to eventually get this directly from APIs, but needs to be scraped from
     websites for now.
-
     """
 
     name: str
-    prompt_token_cost: float
-    completion_token_cost: float
     max_tokens: int
+    prompt_token_cost: float
 
 
 @dataclass
-class ChatModelInfo(ModelInfo):
+class CompletionModelInfo(ModelInfo):
+    """Struct for generic completion model information."""
+
+    completion_token_cost: float
+
+
+@dataclass
+class ChatModelInfo(CompletionModelInfo):
     """Struct for chat model information."""
 
-    pass
+
+@dataclass
+class TextModelInfo(CompletionModelInfo):
+    """Struct for text completion model information."""
 
 
 @dataclass
@@ -38,6 +62,73 @@ class EmbeddingModelInfo(ModelInfo):
     embedding_dimensions: int
 
 
+@dataclass
+class ChatSequence:
+    """Utility container for a chat sequence"""
+
+    model: ChatModelInfo
+    messages: list[Message] = field(default_factory=list)
+
+    def __getitem__(self, i: int):
+        return self.messages[i]
+
+    def __iter__(self):
+        return iter(self.messages)
+
+    def __len__(self):
+        return len(self.messages)
+
+    def append(self, message: Message):
+        return self.messages.append(message)
+
+    def extend(self, messages: list[Message] | ChatSequence):
+        return self.messages.extend(messages)
+
+    def insert(self, index: int, *messages: Message):
+        for message in reversed(messages):
+            self.messages.insert(index, message)
+
+    @classmethod
+    def for_model(cls, model_name: str, messages: list[Message] | ChatSequence = []):
+        from autogpt.llm.providers.openai import OPEN_AI_CHAT_MODELS
+
+        if not model_name in OPEN_AI_CHAT_MODELS:
+            raise ValueError(f"Unknown chat model '{model_name}'")
+
+        return ChatSequence(
+            model=OPEN_AI_CHAT_MODELS[model_name], messages=list(messages)
+        )
+
+    def add(self, message_role: MessageRole, content: str):
+        self.messages.append(Message(message_role, content))
+
+    @property
+    def token_length(self):
+        from autogpt.llm.utils import count_message_tokens
+
+        return count_message_tokens(self.messages, self.model.name)
+
+    def raw(self) -> list[MessageDict]:
+        return [m.raw() for m in self.messages]
+
+    def dump(self) -> str:
+        SEPARATOR_LENGTH = 42
+
+        def separator(text: str):
+            half_sep_len = (SEPARATOR_LENGTH - 2 - len(text)) / 2
+            return f"{floor(half_sep_len)*'-'} {text.upper()} {ceil(half_sep_len)*'-'}"
+
+        formatted_messages = "\n".join(
+            [f"{separator(m.role)}\n{m.content}" for m in self.messages]
+        )
+        return f"""
+============== ChatSequence ==============
+Length: {self.token_length} tokens; {len(self.messages)} messages
+{formatted_messages}
+==========================================
+"""
+
+
 @dataclass
 class LLMResponse:
     """Standard response struct for a response from an LLM model."""

autogpt/llm/chat.py

@@ -1,253 +1,202 @@
-import time
-from random import shuffle
+from __future__ import annotations
 
-from openai.error import RateLimitError
+import time
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from autogpt.agent.agent import Agent
 
 from autogpt.config import Config
 from autogpt.llm.api_manager import ApiManager
-from autogpt.llm.base import Message
-from autogpt.llm.llm_utils import create_chat_completion
-from autogpt.llm.token_counter import count_message_tokens
+from autogpt.llm.base import ChatSequence, Message
+from autogpt.llm.utils import count_message_tokens, create_chat_completion
+from autogpt.log_cycle.log_cycle import CURRENT_CONTEXT_FILE_NAME
 from autogpt.logs import logger
-from autogpt.memory_management.store_memory import (
-    save_memory_trimmed_from_context_window,
-)
-from autogpt.memory_management.summary_memory import (
-    get_newly_trimmed_messages,
-    update_running_summary,
-)
-
-cfg = Config()
-
-
-def create_chat_message(role, content) -> Message:
-    """
-    Create a chat message with the given role and content.
-
-    Args:
-    role (str): The role of the message sender, e.g., "system", "user", or "assistant".
-    content (str): The content of the message.
-
-    Returns:
-    dict: A dictionary containing the role and content of the message.
-    """
-    return {"role": role, "content": content}
-
-
-def generate_context(prompt, relevant_memory, full_message_history, model):
-    current_context = [
-        create_chat_message("system", prompt),
-        create_chat_message(
-            "system", f"The current time and date is {time.strftime('%c')}"
-        ),
-        # create_chat_message(
-        #     "system",
-        #     f"This reminds you of these events from your past:\n{relevant_memory}\n\n",
-        # ),
-    ]
-
-    # Add messages from the full message history until we reach the token limit
-    next_message_to_add_index = len(full_message_history) - 1
-    insertion_index = len(current_context)
-    # Count the currently used tokens
-    current_tokens_used = count_message_tokens(current_context, model)
-    return (
-        next_message_to_add_index,
-        current_tokens_used,
-        insertion_index,
-        current_context,
-    )
 
 
 # TODO: Change debug from hardcode to argument
 def chat_with_ai(
-    agent, prompt, user_input, full_message_history, permanent_memory, token_limit
+    config: Config,
+    agent: Agent,
+    system_prompt: str,
+    user_input: str,
+    token_limit: int,
+    model: str | None = None,
 ):
-    """Interact with the OpenAI API, sending the prompt, user input, message history,
-    and permanent memory."""
-    while True:
-        try:
-            """
-            Interact with the OpenAI API, sending the prompt, user input,
-                message history, and permanent memory.
+    """
+    Interact with the OpenAI API, sending the prompt, user input,
+        message history, and permanent memory.
 
-            Args:
-                prompt (str): The prompt explaining the rules to the AI.
-                user_input (str): The input from the user.
-                full_message_history (list): The list of all messages sent between the
-                    user and the AI.
-                permanent_memory (Obj): The memory object containing the permanent
-                  memory.
-                token_limit (int): The maximum number of tokens allowed in the API call.
+    Args:
+        config (Config): The config to use.
+        agent (Agent): The agent to use.
+        system_prompt (str): The prompt explaining the rules to the AI.
+        user_input (str): The input from the user.
+        token_limit (int): The maximum number of tokens allowed in the API call.
+        model (str, optional): The model to use. If None, the config.fast_llm_model will be used. Defaults to None.
 
-            Returns:
-            str: The AI's response.
-            """
-            model = cfg.fast_llm_model  # TODO: Change model from hardcode to argument
-            # Reserve 1000 tokens for the response
-            logger.debug(f"Token limit: {token_limit}")
-            send_token_limit = token_limit - 1000
+    Returns:
+    str: The AI's response.
+    """
+    if model is None:
+        model = config.fast_llm_model
 
-            # if len(full_message_history) == 0:
-            #     relevant_memory = ""
-            # else:
-            #     recent_history = full_message_history[-5:]
-            #     shuffle(recent_history)
-            #     relevant_memories = permanent_memory.get_relevant(
-            #         str(recent_history), 5
-            #     )
-            #     if relevant_memories:
-            #         shuffle(relevant_memories)
-            #     relevant_memory = str(relevant_memories)
-            relevant_memory = ""
-            logger.debug(f"Memory Stats: {permanent_memory.get_stats()}")
+    # Reserve 1000 tokens for the response
+    logger.debug(f"Token limit: {token_limit}")
+    send_token_limit = token_limit - 1000
 
-            (
-                next_message_to_add_index,
-                current_tokens_used,
-                insertion_index,
-                current_context,
-            ) = generate_context(prompt, relevant_memory, full_message_history, model)
+    # if len(agent.history) == 0:
+    #     relevant_memory = ""
+    # else:
+    #     recent_history = agent.history[-5:]
+    #     shuffle(recent_history)
+    #     relevant_memories = agent.memory.get_relevant(
+    #         str(recent_history), 5
+    #     )
+    #     if relevant_memories:
+    #         shuffle(relevant_memories)
+    #     relevant_memory = str(relevant_memories)
+    # logger.debug(f"Memory Stats: {agent.memory.get_stats()}")
+    relevant_memory = []
 
-            # while current_tokens_used > 2500:
-            #     # remove memories until we are under 2500 tokens
-            #     relevant_memory = relevant_memory[:-1]
-            #     (
-            #         next_message_to_add_index,
-            #         current_tokens_used,
-            #         insertion_index,
-            #         current_context,
-            #     ) = generate_context(
-            #         prompt, relevant_memory, full_message_history, model
-            #     )
+    message_sequence = ChatSequence.for_model(
+        model,
+        [
+            Message("system", system_prompt),
+            Message("system", f"The current time and date is {time.strftime('%c')}"),
+            # Message(
+            #     "system",
+            #     f"This reminds you of these events from your past:\n{relevant_memory}\n\n",
+            # ),
+        ],
+    )
 
-            current_tokens_used += count_message_tokens(
-                [create_chat_message("user", user_input)], model
-            )  # Account for user input (appended later)
+    # Add messages from the full message history until we reach the token limit
+    next_message_to_add_index = len(agent.history) - 1
+    insertion_index = len(message_sequence)
+    # Count the currently used tokens
+    current_tokens_used = message_sequence.token_length
 
-            current_tokens_used += 500  # Account for memory (appended later) TODO: The final memory may be less than 500 tokens
+    # while current_tokens_used > 2500:
+    #     # remove memories until we are under 2500 tokens
+    #     relevant_memory = relevant_memory[:-1]
+    #     (
+    #         next_message_to_add_index,
+    #         current_tokens_used,
+    #         insertion_index,
+    #         current_context,
+    #     ) = generate_context(
+    #         prompt, relevant_memory, agent.history, model
+    #     )
 
-            # Add Messages until the token limit is reached or there are no more messages to add.
-            while next_message_to_add_index >= 0:
-                # print (f"CURRENT TOKENS USED: {current_tokens_used}")
-                message_to_add = full_message_history[next_message_to_add_index]
+    # Account for user input (appended later)
+    user_input_msg = Message("user", user_input)
+    current_tokens_used += count_message_tokens([user_input_msg], model)
 
-                tokens_to_add = count_message_tokens([message_to_add], model)
-                if current_tokens_used + tokens_to_add > send_token_limit:
-                    # save_memory_trimmed_from_context_window(
-                    #     full_message_history,
-                    #     next_message_to_add_index,
-                    #     permanent_memory,
-                    # )
-                    break
+    current_tokens_used += 500  # Reserve space for new_summary_message
 
-                # Add the most recent message to the start of the current context,
-                #  after the two system prompts.
-                current_context.insert(
-                    insertion_index, full_message_history[next_message_to_add_index]
-                )
+    # Add Messages until the token limit is reached or there are no more messages to add.
+    for cycle in reversed(list(agent.history.per_cycle())):
+        messages_to_add = [msg for msg in cycle if msg is not None]
+        tokens_to_add = count_message_tokens(messages_to_add, model)
+        if current_tokens_used + tokens_to_add > send_token_limit:
+            break
 
-                # Count the currently used tokens
-                current_tokens_used += tokens_to_add
+        # Add the most recent message to the start of the chain,
+        #  after the system prompts.
+        message_sequence.insert(insertion_index, *messages_to_add)
+        current_tokens_used += tokens_to_add
 
-                # Move to the next most recent message in the full message history
-                next_message_to_add_index -= 1
+    # Update & add summary of trimmed messages
+    if len(agent.history) > 0:
+        new_summary_message, trimmed_messages = agent.history.trim_messages(
+            current_message_chain=list(message_sequence),
+        )
+        tokens_to_add = count_message_tokens([new_summary_message], model)
+        message_sequence.insert(insertion_index, new_summary_message)
+        current_tokens_used += tokens_to_add - 500
 
-            # Insert Memories
-            if len(full_message_history) > 0:
-                (
-                    newly_trimmed_messages,
-                    agent.last_memory_index,
-                ) = get_newly_trimmed_messages(
-                    full_message_history=full_message_history,
-                    current_context=current_context,
-                    last_memory_index=agent.last_memory_index,
-                )
-                agent.summary_memory = update_running_summary(
-                    current_memory=agent.summary_memory,
-                    new_events=newly_trimmed_messages,
-                )
-                current_context.insert(insertion_index, agent.summary_memory)
+        # FIXME: uncomment when memory is back in use
+        # memory_store = get_memory(cfg)
+        # for _, ai_msg, result_msg in agent.history.per_cycle(trimmed_messages):
+        #     memory_to_add = MemoryItem.from_ai_action(ai_msg, result_msg)
+        #     logger.debug(f"Storing the following memory:\n{memory_to_add.dump()}")
+        #     memory_store.add(memory_to_add)
 
-            api_manager = ApiManager()
-            # inform the AI about its remaining budget (if it has one)
-            if api_manager.get_total_budget() > 0.0:
-                remaining_budget = (
-                    api_manager.get_total_budget() - api_manager.get_total_cost()
-                )
-                if remaining_budget < 0:
-                    remaining_budget = 0
-                system_message = (
-                    f"Your remaining API budget is ${remaining_budget:.3f}"
-                    + (
-                        " BUDGET EXCEEDED! SHUT DOWN!\n\n"
-                        if remaining_budget == 0
-                        else " Budget very nearly exceeded! Shut down gracefully!\n\n"
-                        if remaining_budget < 0.005
-                        else " Budget nearly exceeded. Finish up.\n\n"
-                        if remaining_budget < 0.01
-                        else "\n\n"
-                    )
-                )
-                logger.debug(system_message)
-                current_context.append(create_chat_message("system", system_message))
+    api_manager = ApiManager()
+    # inform the AI about its remaining budget (if it has one)
+    if api_manager.get_total_budget() > 0.0:
+        remaining_budget = api_manager.get_total_budget() - api_manager.get_total_cost()
+        if remaining_budget < 0:
+            remaining_budget = 0
+        budget_message = f"Your remaining API budget is ${remaining_budget:.3f}" + (
+            " BUDGET EXCEEDED! SHUT DOWN!\n\n"
+            if remaining_budget == 0
+            else " Budget very nearly exceeded! Shut down gracefully!\n\n"
+            if remaining_budget < 0.005
+            else " Budget nearly exceeded. Finish up.\n\n"
+            if remaining_budget < 0.01
+            else "\n\n"
+        )
+        logger.debug(budget_message)
+        message_sequence.add("system", budget_message)
+        current_tokens_used += count_message_tokens([message_sequence[-1]], model)
 
-            # Append user input, the length of this is accounted for above
-            current_context.extend([create_chat_message("user", user_input)])
+    # Append user input, the length of this is accounted for above
+    message_sequence.append(user_input_msg)
 
-            plugin_count = len(cfg.plugins)
-            for i, plugin in enumerate(cfg.plugins):
-                if not plugin.can_handle_on_planning():
-                    continue
-                plugin_response = plugin.on_planning(
-                    agent.prompt_generator, current_context
-                )
-                if not plugin_response or plugin_response == "":
-                    continue
-                tokens_to_add = count_message_tokens(
-                    [create_chat_message("system", plugin_response)], model
-                )
-                if current_tokens_used + tokens_to_add > send_token_limit:
-                    logger.debug("Plugin response too long, skipping:", plugin_response)
-                    logger.debug("Plugins remaining at stop:", plugin_count - i)
-                    break
-                current_context.append(create_chat_message("system", plugin_response))
+    plugin_count = len(config.plugins)
+    for i, plugin in enumerate(config.plugins):
+        if not plugin.can_handle_on_planning():
+            continue
+        plugin_response = plugin.on_planning(
+            agent.ai_config.prompt_generator, message_sequence.raw()
+        )
+        if not plugin_response or plugin_response == "":
+            continue
+        tokens_to_add = count_message_tokens(
+            [Message("system", plugin_response)], model
+        )
+        if current_tokens_used + tokens_to_add > send_token_limit:
+            logger.debug(f"Plugin response too long, skipping: {plugin_response}")
+            logger.debug(f"Plugins remaining at stop: {plugin_count - i}")
+            break
+        message_sequence.add("system", plugin_response)
+    # Calculate remaining tokens
+    tokens_remaining = token_limit - current_tokens_used
+    # assert tokens_remaining >= 0, "Tokens remaining is negative.
+    # This should never happen, please submit a bug report at
+    #  https://www.github.com/Torantulino/Auto-GPT"
 
-            # Calculate remaining tokens
-            tokens_remaining = token_limit - current_tokens_used
-            # assert tokens_remaining >= 0, "Tokens remaining is negative.
-            # This should never happen, please submit a bug report at
-            #  https://www.github.com/Torantulino/Auto-GPT"
+    # Debug print the current context
+    logger.debug(f"Token limit: {token_limit}")
+    logger.debug(f"Send Token Count: {current_tokens_used}")
+    logger.debug(f"Tokens remaining for response: {tokens_remaining}")
+    logger.debug("------------ CONTEXT SENT TO AI ---------------")
+    for message in message_sequence:
+        # Skip printing the prompt
+        if message.role == "system" and message.content == system_prompt:
+            continue
+        logger.debug(f"{message.role.capitalize()}: {message.content}")
+        logger.debug("")
+    logger.debug("----------- END OF CONTEXT ----------------")
+    agent.log_cycle_handler.log_cycle(
+        agent.ai_name,
+        agent.created_at,
+        agent.cycle_count,
+        message_sequence.raw(),
+        CURRENT_CONTEXT_FILE_NAME,
+    )
 
-            # Debug print the current context
-            logger.debug(f"Token limit: {token_limit}")
-            logger.debug(f"Send Token Count: {current_tokens_used}")
-            logger.debug(f"Tokens remaining for response: {tokens_remaining}")
-            logger.debug("------------ CONTEXT SENT TO AI ---------------")
-            for message in current_context:
-                # Skip printing the prompt
-                if message["role"] == "system" and message["content"] == prompt:
-                    continue
-                logger.debug(f"{message['role'].capitalize()}: {message['content']}")
-                logger.debug("")
-            logger.debug("----------- END OF CONTEXT ----------------")
+    # TODO: use a model defined elsewhere, so that model can contain
+    # temperature and other settings we care about
+    assistant_reply = create_chat_completion(
+        prompt=message_sequence,
+        max_tokens=tokens_remaining,
+    )
 
-            # TODO: use a model defined elsewhere, so that model can contain
-            # temperature and other settings we care about
-            assistant_reply = create_chat_completion(
-                model=model,
-                messages=current_context,
-                max_tokens=tokens_remaining,
-            )
+    # Update full message history
+    agent.history.append(user_input_msg)
+    agent.history.add("assistant", assistant_reply, "ai_response")
 
-            # Update full message history
-            full_message_history.append(create_chat_message("user", user_input))
-            full_message_history.append(
-                create_chat_message("assistant", assistant_reply)
-            )
-
-            return assistant_reply
-        except RateLimitError:
-            # TODO: When we switch to langchain, this is built in
-            logger.warn("Error: ", "API Rate Limit Reached. Waiting 10 seconds...")
-            time.sleep(10)
+    return assistant_reply

autogpt/llm/modelsinfo.py

@@ -1,7 +0,0 @@
-COSTS = {
-    "gpt-3.5-turbo": {"prompt": 0.002, "completion": 0.002},
-    "gpt-3.5-turbo-0301": {"prompt": 0.002, "completion": 0.002},
-    "gpt-4-0314": {"prompt": 0.03, "completion": 0.06},
-    "gpt-4": {"prompt": 0.03, "completion": 0.06},
-    "text-embedding-ada-002": {"prompt": 0.0004, "completion": 0.0},
-}

autogpt/llm/providers/openai.py

@@ -1,37 +1,90 @@
-from autogpt.llm.base import ChatModelInfo, EmbeddingModelInfo
+from autogpt.llm.base import ChatModelInfo, EmbeddingModelInfo, TextModelInfo
 
 OPEN_AI_CHAT_MODELS = {
-    "gpt-3.5-turbo": ChatModelInfo(
-        name="gpt-3.5-turbo",
-        prompt_token_cost=0.002,
-        completion_token_cost=0.002,
-        max_tokens=4096,
-    ),
-    "gpt-4": ChatModelInfo(
-        name="gpt-4",
-        prompt_token_cost=0.03,
-        completion_token_cost=0.06,
-        max_tokens=8192,
-    ),
-    "gpt-4-32k": ChatModelInfo(
-        name="gpt-4-32k",
-        prompt_token_cost=0.06,
-        completion_token_cost=0.12,
-        max_tokens=32768,
-    ),
+    info.name: info
+    for info in [
+        ChatModelInfo(
+            name="gpt-3.5-turbo-0301",
+            prompt_token_cost=0.0015,
+            completion_token_cost=0.002,
+            max_tokens=4096,
+        ),
+        ChatModelInfo(
+            name="gpt-3.5-turbo-0613",
+            prompt_token_cost=0.0015,
+            completion_token_cost=0.002,
+            max_tokens=4096,
+        ),
+        ChatModelInfo(
+            name="gpt-3.5-turbo-16k-0613",
+            prompt_token_cost=0.003,
+            completion_token_cost=0.004,
+            max_tokens=16384,
+        ),
+        ChatModelInfo(
+            name="gpt-4-0314",
+            prompt_token_cost=0.03,
+            completion_token_cost=0.06,
+            max_tokens=8192,
+        ),
+        ChatModelInfo(
+            name="gpt-4-0613",
+            prompt_token_cost=0.03,
+            completion_token_cost=0.06,
+            max_tokens=8192,
+        ),
+        ChatModelInfo(
+            name="gpt-4-32k-0314",
+            prompt_token_cost=0.06,
+            completion_token_cost=0.12,
+            max_tokens=32768,
+        ),
+        ChatModelInfo(
+            name="gpt-4-32k-0613",
+            prompt_token_cost=0.06,
+            completion_token_cost=0.12,
+            max_tokens=32768,
+        ),
+    ]
+}
+# Set aliases for rolling model IDs
+chat_model_mapping = {
+    "gpt-3.5-turbo": "gpt-3.5-turbo-0301",
+    "gpt-3.5-turbo-16k": "gpt-3.5-turbo-16k-0613",
+    "gpt-4": "gpt-4-0314",
+    "gpt-4-32k": "gpt-4-32k-0314",
+}
+for alias, target in chat_model_mapping.items():
+    alias_info = ChatModelInfo(**OPEN_AI_CHAT_MODELS[target].__dict__)
+    alias_info.name = alias
+    OPEN_AI_CHAT_MODELS[alias] = alias_info
+
+OPEN_AI_TEXT_MODELS = {
+    info.name: info
+    for info in [
+        TextModelInfo(
+            name="text-davinci-003",
+            prompt_token_cost=0.02,
+            completion_token_cost=0.02,
+            max_tokens=4097,
+        ),
+    ]
 }
 
 OPEN_AI_EMBEDDING_MODELS = {
-    "text-embedding-ada-002": EmbeddingModelInfo(
-        name="text-embedding-ada-002",
-        prompt_token_cost=0.0004,
-        completion_token_cost=0.0,
-        max_tokens=8191,
-        embedding_dimensions=1536,
-    ),
+    info.name: info
+    for info in [
+        EmbeddingModelInfo(
+            name="text-embedding-ada-002",
+            prompt_token_cost=0.0001,
+            max_tokens=8191,
+            embedding_dimensions=1536,
+        ),
+    ]
 }
 
-OPEN_AI_MODELS = {
+OPEN_AI_MODELS: dict[str, ChatModelInfo | EmbeddingModelInfo | TextModelInfo] = {
     **OPEN_AI_CHAT_MODELS,
+    **OPEN_AI_TEXT_MODELS,
     **OPEN_AI_EMBEDDING_MODELS,
 }

autogpt/llm/utils/__init__.py

@@ -2,17 +2,59 @@ from __future__ import annotations
 
 import functools
 import time
-from typing import List, Optional
+from typing import List, Literal, Optional
+from unittest.mock import patch
 
 import openai
+import openai.api_resources.abstract.engine_api_resource as engine_api_resource
+import openai.util
 from colorama import Fore, Style
-from openai.error import APIError, RateLimitError, Timeout
+from openai.error import APIError, RateLimitError
+from openai.openai_object import OpenAIObject
 
 from autogpt.config import Config
-from autogpt.llm.api_manager import ApiManager
-from autogpt.llm.base import Message
 from autogpt.logs import logger
 
+from ..api_manager import ApiManager
+from ..base import ChatSequence, Message
+from ..providers.openai import OPEN_AI_CHAT_MODELS
+from .token_counter import *
+
+
+def metered(func):
+    """Adds ApiManager metering to functions which make OpenAI API calls"""
+    api_manager = ApiManager()
+
+    openai_obj_processor = openai.util.convert_to_openai_object
+
+    def update_usage_with_response(response: OpenAIObject):
+        try:
+            usage = response.usage
+            logger.debug(f"Reported usage from call to model {response.model}: {usage}")
+            api_manager.update_cost(
+                response.usage.prompt_tokens,
+                response.usage.completion_tokens if "completion_tokens" in usage else 0,
+                response.model,
+            )
+        except Exception as err:
+            logger.warn(f"Failed to update API costs: {err.__class__.__name__}: {err}")
+
+    def metering_wrapper(*args, **kwargs):
+        openai_obj = openai_obj_processor(*args, **kwargs)
+        if isinstance(openai_obj, OpenAIObject) and "usage" in openai_obj:
+            update_usage_with_response(openai_obj)
+        return openai_obj
+
+    def metered_func(*args, **kwargs):
+        with patch.object(
+            engine_api_resource.util,
+            "convert_to_openai_object",
+            side_effect=metering_wrapper,
+        ):
+            return func(*args, **kwargs)
+
+    return metered_func
+
 
 def retry_openai_api(
     num_retries: int = 10,
@@ -30,7 +72,7 @@ def retry_openai_api(
     api_key_error_msg = (
         f"Please double check that you have setup a "
         f"{Fore.CYAN + Style.BRIGHT}PAID{Style.RESET_ALL} OpenAI API Account. You can "
-        f"read more here: {Fore.CYAN}https://significant-gravitas.github.io/Auto-GPT/setup/#getting-an-api-key{Fore.RESET}"
+        f"read more here: {Fore.CYAN}https://docs.agpt.co/setup/#getting-an-api-key{Fore.RESET}"
     )
     backoff_msg = (
         f"{Fore.RED}Error: API Bad gateway. Waiting {{backoff}} seconds...{Fore.RESET}"
@@ -55,7 +97,7 @@ def retry_openai_api(
                         user_warned = True
 
                 except APIError as e:
-                    if (e.http_status != 502) or (attempt == num_attempts):
+                    if (e.http_status not in [502, 429]) or (attempt == num_attempts):
                         raise
 
                 backoff = backoff_base ** (attempt + 2)
@@ -68,7 +110,11 @@ def retry_openai_api(
 
 
 def call_ai_function(
-    function: str, args: list, description: str, model: str | None = None
+    function: str,
+    args: list,
+    description: str,
+    model: str | None = None,
+    config: Config = None,
 ) -> str:
     """Call an AI function
 
@@ -84,29 +130,62 @@ def call_ai_function(
     Returns:
         str: The response from the function
     """
-    cfg = Config()
     if model is None:
-        model = cfg.smart_llm_model
+        model = config.smart_llm_model
     # For each arg, if any are None, convert to "None":
     args = [str(arg) if arg is not None else "None" for arg in args]
     # parse args to comma separated string
-    args: str = ", ".join(args)
-    messages: List[Message] = [
-        {
-            "role": "system",
-            "content": f"You are now the following python function: ```# {description}"
-            f"\n{function}```\n\nOnly respond with your `return` value.",
-        },
-        {"role": "user", "content": args},
-    ]
+    arg_str: str = ", ".join(args)
 
-    return create_chat_completion(model=model, messages=messages, temperature=0)
+    prompt = ChatSequence.for_model(
+        model,
+        [
+            Message(
+                "system",
+                f"You are now the following python function: ```# {description}"
+                f"\n{function}```\n\nOnly respond with your `return` value.",
+            ),
+            Message("user", arg_str),
+        ],
+    )
+    return create_chat_completion(prompt=prompt, temperature=0)
+
+
+@metered
+@retry_openai_api()
+def create_text_completion(
+    prompt: str,
+    model: Optional[str],
+    temperature: Optional[float],
+    max_output_tokens: Optional[int],
+) -> str:
+    cfg = Config()
+    if model is None:
+        model = cfg.fast_llm_model
+    if temperature is None:
+        temperature = cfg.temperature
+
+    if cfg.use_azure:
+        kwargs = {"deployment_id": cfg.get_azure_deployment_id_for_model(model)}
+    else:
+        kwargs = {"model": model}
+
+    response = openai.Completion.create(
+        **kwargs,
+        prompt=prompt,
+        temperature=temperature,
+        max_tokens=max_output_tokens,
+        api_key=cfg.openai_api_key,
+    )
+    return response.choices[0].text
 
 
 # Overly simple abstraction until we create something better
 # simple retry mechanism when getting a rate error or a bad gateway
+@metered
+@retry_openai_api()
 def create_chat_completion(
-    messages: List[Message],  # type: ignore
+    prompt: ChatSequence,
     model: Optional[str] = None,
     temperature: float = None,
     max_tokens: Optional[int] = None,
@@ -123,23 +202,25 @@ def create_chat_completion(
         str: The response from the chat completion
     """
     cfg = Config()
+    if model is None:
+        model = prompt.model.name
     if temperature is None:
         temperature = cfg.temperature
+    if max_tokens is None:
+        max_tokens = OPEN_AI_CHAT_MODELS[model].max_tokens - prompt.token_length
 
-    num_retries = 10
-    warned_user = False
     logger.debug(
         f"{Fore.GREEN}Creating chat completion with model {model}, temperature {temperature}, max_tokens {max_tokens}{Fore.RESET}"
     )
     for plugin in cfg.plugins:
         if plugin.can_handle_chat_completion(
-            messages=messages,
+            messages=prompt.raw(),
             model=model,
             temperature=temperature,
             max_tokens=max_tokens,
         ):
             message = plugin.handle_chat_completion(
-                messages=messages,
+                messages=prompt.raw(),
                 model=model,
                 temperature=temperature,
                 max_tokens=max_tokens,
@@ -148,58 +229,20 @@ def create_chat_completion(
                 return message
     api_manager = ApiManager()
     response = None
-    for attempt in range(num_retries):
-        backoff = 2 ** (attempt + 2)
-        try:
-            if cfg.use_azure:
-                response = api_manager.create_chat_completion(
-                    deployment_id=cfg.get_azure_deployment_id_for_model(model),
-                    model=model,
-                    messages=messages,
-                    temperature=temperature,
-                    max_tokens=max_tokens,
-                )
-            else:
-                response = api_manager.create_chat_completion(
-                    model=model,
-                    messages=messages,
-                    temperature=temperature,
-                    max_tokens=max_tokens,
-                )
-            break
-        except RateLimitError:
-            logger.debug(
-                f"{Fore.RED}Error: ", f"Reached rate limit, passing...{Fore.RESET}"
-            )
-            if not warned_user:
-                logger.double_check(
-                    f"Please double check that you have setup a {Fore.CYAN + Style.BRIGHT}PAID{Style.RESET_ALL} OpenAI API Account. "
-                    + f"You can read more here: {Fore.CYAN}https://significant-gravitas.github.io/Auto-GPT/setup/#getting-an-api-key{Fore.RESET}"
-                )
-                warned_user = True
-        except (APIError, Timeout) as e:
-            if e.http_status != 502:
-                raise
-            if attempt == num_retries - 1:
-                raise
-        logger.debug(
-            f"{Fore.RED}Error: ",
-            f"API Bad gateway. Waiting {backoff} seconds...{Fore.RESET}",
-        )
-        time.sleep(backoff)
-    if response is None:
-        logger.typewriter_log(
-            "FAILED TO GET RESPONSE FROM OPENAI",
-            Fore.RED,
-            "Auto-GPT has failed to get a response from OpenAI's services. "
-            + f"Try running Auto-GPT again, and if the problem the persists try running it with `{Fore.CYAN}--debug{Fore.RESET}`.",
-        )
-        logger.double_check()
-        if cfg.debug_mode:
-            raise RuntimeError(f"Failed to get response after {num_retries} retries")
-        else:
-            quit(1)
-    resp = response.choices[0].message["content"]
+
+    if cfg.use_azure:
+        kwargs = {"deployment_id": cfg.get_azure_deployment_id_for_model(model)}
+    else:
+        kwargs = {"model": model}
+
+    response = api_manager.create_chat_completion(
+        **kwargs,
+        messages=prompt.raw(),
+        temperature=temperature,
+        max_tokens=max_tokens,
+    )
+
+    resp = response.choices[0].message.content
     for plugin in cfg.plugins:
         if not plugin.can_handle_on_response():
             continue
@@ -207,52 +250,20 @@ def create_chat_completion(
     return resp
 
 
-def get_ada_embedding(text: str) -> List[float]:
-    """Get an embedding from the ada model.
-
-    Args:
-        text (str): The text to embed.
-
-    Returns:
-        List[float]: The embedding.
-    """
-    cfg = Config()
-    model = "text-embedding-ada-002"
-    text = text.replace("\n", " ")
-
-    if cfg.use_azure:
-        kwargs = {"engine": cfg.get_azure_deployment_id_for_model(model)}
-    else:
-        kwargs = {"model": model}
-
-    embedding = create_embedding(text, **kwargs)
+def check_model(
+    model_name: str, model_type: Literal["smart_llm_model", "fast_llm_model"]
+) -> str:
+    """Check if model is available for use. If not, return gpt-3.5-turbo."""
     api_manager = ApiManager()
-    api_manager.update_cost(
-        prompt_tokens=embedding.usage.prompt_tokens,
-        completion_tokens=0,
-        model=model,
-    )
-    return embedding["data"][0]["embedding"]
-
-
-@retry_openai_api()
-def create_embedding(
-    text: str,
-    *_,
-    **kwargs,
-) -> openai.Embedding:
-    """Create an embedding using the OpenAI API
-
-    Args:
-        text (str): The text to embed.
-        kwargs: Other arguments to pass to the OpenAI API embedding creation call.
-
-    Returns:
-        openai.Embedding: The embedding object.
-    """
-    cfg = Config()
-    return openai.Embedding.create(
-        input=[text],
-        api_key=cfg.openai_api_key,
-        **kwargs,
+    models = api_manager.get_models()
+
+    if any(model_name in m["id"] for m in models):
+        return model_name
+
+    logger.typewriter_log(
+        "WARNING: ",
+        Fore.YELLOW,
+        f"You do not have access to {model_name}. Setting {model_type} to "
+        f"gpt-3.5-turbo.",
     )
+    return "gpt-3.5-turbo"

autogpt/llm/utils/token_counter.py

@@ -24,36 +24,32 @@ def count_message_tokens(
     Returns:
         int: The number of tokens used by the list of messages.
     """
-    try:
-        encoding = tiktoken.encoding_for_model(model)
-    except KeyError:
-        logger.warn("Warning: model not found. Using cl100k_base encoding.")
-        encoding = tiktoken.get_encoding("cl100k_base")
-    if model == "gpt-3.5-turbo":
-        # !Note: gpt-3.5-turbo may change over time.
-        # Returning num tokens assuming gpt-3.5-turbo-0301.")
-        return count_message_tokens(messages, model="gpt-3.5-turbo-0301")
-    elif model == "gpt-4":
-        # !Note: gpt-4 may change over time. Returning num tokens assuming gpt-4-0314.")
-        return count_message_tokens(messages, model="gpt-4-0314")
-    elif model == "gpt-3.5-turbo-0301":
+    if model.startswith("gpt-3.5-turbo"):
         tokens_per_message = (
             4  # every message follows <|start|>{role/name}\n{content}<|end|>\n
         )
         tokens_per_name = -1  # if there's a name, the role is omitted
-    elif model == "gpt-4-0314":
+        encoding_model = "gpt-3.5-turbo"
+    elif model.startswith("gpt-4"):
         tokens_per_message = 3
         tokens_per_name = 1
+        encoding_model = "gpt-4"
     else:
         raise NotImplementedError(
-            f"num_tokens_from_messages() is not implemented for model {model}.\n"
+            f"count_message_tokens() is not implemented for model {model}.\n"
             " See https://github.com/openai/openai-python/blob/main/chatml.md for"
             " information on how messages are converted to tokens."
         )
+    try:
+        encoding = tiktoken.encoding_for_model(encoding_model)
+    except KeyError:
+        logger.warn("Warning: model not found. Using cl100k_base encoding.")
+        encoding = tiktoken.get_encoding("cl100k_base")
+
     num_tokens = 0
     for message in messages:
         num_tokens += tokens_per_message
-        for key, value in message.items():
+        for key, value in message.raw().items():
             num_tokens += len(encoding.encode(value))
             if key == "name":
                 num_tokens += tokens_per_name

autogpt/log_cycle/json_handler.py

@@ -0,0 +1,20 @@
+import json
+import logging
+
+
+class JsonFileHandler(logging.FileHandler):
+    def __init__(self, filename, mode="a", encoding=None, delay=False):
+        super().__init__(filename, mode, encoding, delay)
+
+    def emit(self, record):
+        json_data = json.loads(self.format(record))
+        with open(self.baseFilename, "w", encoding="utf-8") as f:
+            json.dump(json_data, f, ensure_ascii=False, indent=4)
+
+
+import logging
+
+
+class JsonFormatter(logging.Formatter):
+    def format(self, record):
+        return record.msg

autogpt/log_cycle/log_cycle.py

@@ -0,0 +1,88 @@
+import json
+import os
+from typing import Any, Dict, Union
+
+from autogpt.logs import logger
+
+DEFAULT_PREFIX = "agent"
+FULL_MESSAGE_HISTORY_FILE_NAME = "full_message_history.json"
+CURRENT_CONTEXT_FILE_NAME = "current_context.json"
+NEXT_ACTION_FILE_NAME = "next_action.json"
+PROMPT_SUMMARY_FILE_NAME = "prompt_summary.json"
+SUMMARY_FILE_NAME = "summary.txt"
+SUPERVISOR_FEEDBACK_FILE_NAME = "supervisor_feedback.txt"
+PROMPT_SUPERVISOR_FEEDBACK_FILE_NAME = "prompt_supervisor_feedback.json"
+USER_INPUT_FILE_NAME = "user_input.txt"
+
+
+class LogCycleHandler:
+    """
+    A class for logging cycle data.
+    """
+
+    def __init__(self):
+        self.log_count_within_cycle = 0
+
+    @staticmethod
+    def create_directory_if_not_exists(directory_path: str) -> None:
+        if not os.path.exists(directory_path):
+            os.makedirs(directory_path, exist_ok=True)
+
+    def create_outer_directory(self, ai_name: str, created_at: str) -> str:
+        log_directory = logger.get_log_directory()
+
+        if os.environ.get("OVERWRITE_DEBUG") == "1":
+            outer_folder_name = "auto_gpt"
+        else:
+            ai_name_short = self.get_agent_short_name(ai_name)
+            outer_folder_name = f"{created_at}_{ai_name_short}"
+
+        outer_folder_path = os.path.join(log_directory, "DEBUG", outer_folder_name)
+        self.create_directory_if_not_exists(outer_folder_path)
+
+        return outer_folder_path
+
+    def get_agent_short_name(self, ai_name):
+        return ai_name[:15].rstrip() if ai_name else DEFAULT_PREFIX
+
+    def create_inner_directory(self, outer_folder_path: str, cycle_count: int) -> str:
+        nested_folder_name = str(cycle_count).zfill(3)
+        nested_folder_path = os.path.join(outer_folder_path, nested_folder_name)
+        self.create_directory_if_not_exists(nested_folder_path)
+
+        return nested_folder_path
+
+    def create_nested_directory(
+        self, ai_name: str, created_at: str, cycle_count: int
+    ) -> str:
+        outer_folder_path = self.create_outer_directory(ai_name, created_at)
+        nested_folder_path = self.create_inner_directory(outer_folder_path, cycle_count)
+
+        return nested_folder_path
+
+    def log_cycle(
+        self,
+        ai_name: str,
+        created_at: str,
+        cycle_count: int,
+        data: Union[Dict[str, Any], Any],
+        file_name: str,
+    ) -> None:
+        """
+        Log cycle data to a JSON file.
+
+        Args:
+            data (Any): The data to be logged.
+            file_name (str): The name of the file to save the logged data.
+        """
+        nested_folder_path = self.create_nested_directory(
+            ai_name, created_at, cycle_count
+        )
+
+        json_data = json.dumps(data, ensure_ascii=False, indent=4)
+        log_file_path = os.path.join(
+            nested_folder_path, f"{self.log_count_within_cycle}_{file_name}"
+        )
+
+        logger.log_json(json_data, log_file_path)
+        self.log_count_within_cycle += 1

autogpt/logs.py

@@ -5,9 +5,11 @@ import random
 import re
 import time
 from logging import LogRecord
+from typing import Any
 
 from colorama import Fore, Style
 
+from autogpt.log_cycle.json_handler import JsonFileHandler, JsonFormatter
 from autogpt.singleton import Singleton
 from autogpt.speech import say_text
 
@@ -74,7 +76,13 @@ class Logger(metaclass=Singleton):
         self.logger.addHandler(error_handler)
         self.logger.setLevel(logging.DEBUG)
 
+        self.json_logger = logging.getLogger("JSON_LOGGER")
+        self.json_logger.addHandler(self.file_handler)
+        self.json_logger.addHandler(error_handler)
+        self.json_logger.setLevel(logging.DEBUG)
+
         self.speak_mode = False
+        self.chat_plugins = []
 
     def typewriter_log(
         self, title="", title_color="", content="", speak_text=False, level=logging.INFO
@@ -82,6 +90,9 @@ class Logger(metaclass=Singleton):
         if speak_text and self.speak_mode:
             say_text(f"{title}. {content}")
 
+        for plugin in self.chat_plugins:
+            plugin.report(f"{title}. {content}")
+
         if content:
             if isinstance(content, list):
                 content = " ".join(content)
@@ -148,6 +159,26 @@ class Logger(metaclass=Singleton):
 
         self.typewriter_log("DOUBLE CHECK CONFIGURATION", Fore.YELLOW, additionalText)
 
+    def log_json(self, data: Any, file_name: str) -> None:
+        # Define log directory
+        this_files_dir_path = os.path.dirname(__file__)
+        log_dir = os.path.join(this_files_dir_path, "../logs")
+
+        # Create a handler for JSON files
+        json_file_path = os.path.join(log_dir, file_name)
+        json_data_handler = JsonFileHandler(json_file_path)
+        json_data_handler.setFormatter(JsonFormatter())
+
+        # Log the JSON data using the custom file handler
+        self.json_logger.addHandler(json_data_handler)
+        self.json_logger.debug(data)
+        self.json_logger.removeHandler(json_data_handler)
+
+    def get_log_directory(self):
+        this_files_dir_path = os.path.dirname(__file__)
+        log_dir = os.path.join(this_files_dir_path, "../logs")
+        return os.path.abspath(log_dir)
+
 
 """
 Output stream to console using simulated typing
@@ -195,12 +226,16 @@ class AutoGptFormatter(logging.Formatter):
         if hasattr(record, "color"):
             record.title_color = (
                 getattr(record, "color")
-                + getattr(record, "title")
+                + getattr(record, "title", "")
                 + " "
                 + Style.RESET_ALL
             )
         else:
-            record.title_color = getattr(record, "title")
+            record.title_color = getattr(record, "title", "")
+
+        # Add this line to set 'title' to an empty string if it doesn't exist
+        record.title = getattr(record, "title", "")
+
         if hasattr(record, "msg"):
             record.message_no_color = remove_color_codes(getattr(record, "msg"))
         else:
@@ -252,5 +287,8 @@ def print_assistant_thoughts(
             logger.typewriter_log("- ", Fore.GREEN, line.strip())
     logger.typewriter_log("CRITICISM:", Fore.YELLOW, f"{assistant_thoughts_criticism}")
     # Speak the assistant's thoughts
-    if speak_mode and assistant_thoughts_speak:
-        say_text(assistant_thoughts_speak)
+    if assistant_thoughts_speak:
+        if speak_mode:
+            say_text(assistant_thoughts_speak)
+        else:
+            logger.typewriter_log("SPEAK:", Fore.YELLOW, f"{assistant_thoughts_speak}")

autogpt/main.py

@@ -3,25 +3,46 @@ import logging
 import sys
 from pathlib import Path
 
-from colorama import Fore
+from colorama import Fore, Style
 
-from autogpt.agent.agent import Agent
+from autogpt.agent import Agent
 from autogpt.commands.command import CommandRegistry
 from autogpt.config import Config, check_openai_api_key
 from autogpt.configurator import create_config
 from autogpt.logs import logger
-from autogpt.memory import get_memory
+from autogpt.memory.vector import get_memory
 from autogpt.plugins import scan_plugins
 from autogpt.prompts.prompt import DEFAULT_TRIGGERING_PROMPT, construct_main_ai_config
-from autogpt.utils import get_current_git_branch, get_latest_bulletin
+from autogpt.utils import (
+    get_current_git_branch,
+    get_latest_bulletin,
+    get_legal_warning,
+    markdown_to_ansi_style,
+)
 from autogpt.workspace import Workspace
 from scripts.install_plugin_deps import install_plugin_dependencies
 
+COMMAND_CATEGORIES = [
+    "autogpt.commands.analyze_code",
+    "autogpt.commands.audio_text",
+    "autogpt.commands.execute_code",
+    "autogpt.commands.file_operations",
+    "autogpt.commands.git_operations",
+    "autogpt.commands.google_search",
+    "autogpt.commands.image_gen",
+    "autogpt.commands.improve_code",
+    "autogpt.commands.web_selenium",
+    "autogpt.commands.write_tests",
+    "autogpt.app",
+    "autogpt.commands.task_statuses",
+]
+
 
 def run_auto_gpt(
     continuous: bool,
     continuous_limit: int,
     ai_settings: str,
+    prompt_settings: str,
     skip_reprompt: bool,
     speak: bool,
     debug: bool,
@@ -41,10 +62,13 @@ def run_auto_gpt(
     cfg = Config()
     # TODO: fill in llm values here
     check_openai_api_key()
+
     create_config(
+        cfg,
         continuous,
         continuous_limit,
         ai_settings,
+        prompt_settings,
         skip_reprompt,
         speak,
         debug,
@@ -56,10 +80,24 @@ def run_auto_gpt(
         skip_news,
     )
 
+    if cfg.continuous_mode:
+        for line in get_legal_warning().split("\n"):
+            logger.warn(markdown_to_ansi_style(line), "LEGAL:", Fore.RED)
+
     if not cfg.skip_news:
-        motd = get_latest_bulletin()
+        motd, is_new_motd = get_latest_bulletin()
         if motd:
-            logger.typewriter_log("NEWS: ", Fore.GREEN, motd)
+            motd = markdown_to_ansi_style(motd)
+            for motd_line in motd.split("\n"):
+                logger.info(motd_line, "NEWS:", Fore.GREEN)
+            if is_new_motd and not cfg.chat_messages_enabled:
+                input(
+                    Fore.MAGENTA
+                    + Style.BRIGHT
+                    + "NEWS: Bulletin was updated! Press Enter to continue..."
+                    + Style.RESET_ALL
+                )
+
         git_branch = get_current_git_branch()
         if git_branch and git_branch != "stable":
             logger.typewriter_log(
@@ -104,27 +142,37 @@ def run_auto_gpt(
     cfg.set_plugins(scan_plugins(cfg, cfg.debug_mode))
     # Create a CommandRegistry instance and scan default folder
     command_registry = CommandRegistry()
-    command_registry.import_commands("autogpt.commands.analyze_code")
-    command_registry.import_commands("autogpt.commands.audio_text")
-    command_registry.import_commands("autogpt.commands.execute_code")
-    command_registry.import_commands("autogpt.commands.file_operations")
-    command_registry.import_commands("autogpt.commands.git_operations")
-    command_registry.import_commands("autogpt.commands.google_search")
-    command_registry.import_commands("autogpt.commands.image_gen")
-    command_registry.import_commands("autogpt.commands.improve_code")
-    command_registry.import_commands("autogpt.commands.twitter")
-    command_registry.import_commands("autogpt.commands.web_selenium")
-    command_registry.import_commands("autogpt.commands.write_tests")
-    command_registry.import_commands("autogpt.app")
+
+    logger.debug(
+        f"The following command categories are disabled: {cfg.disabled_command_categories}"
+    )
+    enabled_command_categories = [
+        x for x in COMMAND_CATEGORIES if x not in cfg.disabled_command_categories
+    ]
+
+    logger.debug(
+        f"The following command categories are enabled: {enabled_command_categories}"
+    )
+
+    for command_category in enabled_command_categories:
+        command_registry.import_commands(command_category)
 
     ai_name = ""
     ai_config = construct_main_ai_config()
     ai_config.command_registry = command_registry
+    if ai_config.ai_name:
+        ai_name = ai_config.ai_name
     # print(prompt)
     # Initialize variables
-    full_message_history = []
     next_action_count = 0
 
+    # add chat plugins capable of report to logger
+    if cfg.chat_messages_enabled:
+        for plugin in cfg.plugins:
+            if hasattr(plugin, "can_handle_report") and plugin.can_handle_report():
+                logger.info(f"Loaded plugin into logger: {plugin.__class__.__name__}")
+                logger.chat_plugins.append(plugin)
+
     # Initialize memory and make sure it is empty.
     # this is particularly important for indexing and referencing pinecone memory
     memory = get_memory(cfg, init=True)
@@ -139,12 +187,12 @@ def run_auto_gpt(
     agent = Agent(
         ai_name=ai_name,
         memory=memory,
-        full_message_history=full_message_history,
         next_action_count=next_action_count,
         command_registry=command_registry,
-        config=ai_config,
         system_prompt=system_prompt,
         triggering_prompt=DEFAULT_TRIGGERING_PROMPT,
         workspace_directory=workspace_directory,
+        ai_config=ai_config,
+        config=cfg,
     )
     agent.start_interaction_loop()

autogpt/memory/__init__.py

@@ -1,96 +0,0 @@
-from autogpt.logs import logger
-from autogpt.memory.local import LocalCache
-from autogpt.memory.no_memory import NoMemory
-
-# List of supported memory backends
-# Add a backend to this list if the import attempt is successful
-supported_memory = ["local", "no_memory"]
-
-try:
-    from autogpt.memory.redismem import RedisMemory
-
-    supported_memory.append("redis")
-except ImportError:
-    RedisMemory = None
-
-try:
-    from autogpt.memory.pinecone import PineconeMemory
-
-    supported_memory.append("pinecone")
-except ImportError:
-    PineconeMemory = None
-
-try:
-    from autogpt.memory.weaviate import WeaviateMemory
-
-    supported_memory.append("weaviate")
-except ImportError:
-    WeaviateMemory = None
-
-try:
-    from autogpt.memory.milvus import MilvusMemory
-
-    supported_memory.append("milvus")
-except ImportError:
-    MilvusMemory = None
-
-
-def get_memory(cfg, init=False):
-    memory = None
-    if cfg.memory_backend == "pinecone":
-        if not PineconeMemory:
-            logger.warn(
-                "Error: Pinecone is not installed. Please install pinecone"
-                " to use Pinecone as a memory backend."
-            )
-        else:
-            memory = PineconeMemory(cfg)
-            if init:
-                memory.clear()
-    elif cfg.memory_backend == "redis":
-        if not RedisMemory:
-            logger.warn(
-                "Error: Redis is not installed. Please install redis-py to"
-                " use Redis as a memory backend."
-            )
-        else:
-            memory = RedisMemory(cfg)
-    elif cfg.memory_backend == "weaviate":
-        if not WeaviateMemory:
-            logger.warn(
-                "Error: Weaviate is not installed. Please install weaviate-client to"
-                " use Weaviate as a memory backend."
-            )
-        else:
-            memory = WeaviateMemory(cfg)
-    elif cfg.memory_backend == "milvus":
-        if not MilvusMemory:
-            logger.warn(
-                "Error: pymilvus sdk is not installed."
-                "Please install pymilvus to use Milvus or Zilliz Cloud as memory backend."
-            )
-        else:
-            memory = MilvusMemory(cfg)
-    elif cfg.memory_backend == "no_memory":
-        memory = NoMemory(cfg)
-
-    if memory is None:
-        memory = LocalCache(cfg)
-        if init:
-            memory.clear()
-    return memory
-
-
-def get_supported_memory_backends():
-    return supported_memory
-
-
-__all__ = [
-    "get_memory",
-    "LocalCache",
-    "RedisMemory",
-    "PineconeMemory",
-    "NoMemory",
-    "MilvusMemory",
-    "WeaviateMemory",
-]

autogpt/memory/base.py

@@ -1,31 +0,0 @@
-"""Base class for memory providers."""
-import abc
-
-from autogpt.singleton import AbstractSingleton
-
-
-class MemoryProviderSingleton(AbstractSingleton):
-    @abc.abstractmethod
-    def add(self, data):
-        """Adds to memory"""
-        pass
-
-    @abc.abstractmethod
-    def get(self, data):
-        """Gets from memory"""
-        pass
-
-    @abc.abstractmethod
-    def clear(self):
-        """Clears memory"""
-        pass
-
-    @abc.abstractmethod
-    def get_relevant(self, data, num_relevant=5):
-        """Gets relevant memory for"""
-        pass
-
-    @abc.abstractmethod
-    def get_stats(self):
-        """Get stats from memory"""
-        pass

autogpt/memory/local.py

@@ -1,126 +0,0 @@
-from __future__ import annotations
-
-import dataclasses
-from pathlib import Path
-from typing import Any, List
-
-import numpy as np
-import orjson
-
-from autogpt.llm import get_ada_embedding
-from autogpt.memory.base import MemoryProviderSingleton
-
-EMBED_DIM = 1536
-SAVE_OPTIONS = orjson.OPT_SERIALIZE_NUMPY | orjson.OPT_SERIALIZE_DATACLASS
-
-
-def create_default_embeddings():
-    return np.zeros((0, EMBED_DIM)).astype(np.float32)
-
-
-@dataclasses.dataclass
-class CacheContent:
-    texts: List[str] = dataclasses.field(default_factory=list)
-    embeddings: np.ndarray = dataclasses.field(
-        default_factory=create_default_embeddings
-    )
-
-
-class LocalCache(MemoryProviderSingleton):
-    """A class that stores the memory in a local file"""
-
-    def __init__(self, cfg) -> None:
-        """Initialize a class instance
-
-        Args:
-            cfg: Config object
-
-        Returns:
-            None
-        """
-        workspace_path = Path(cfg.workspace_path)
-        self.filename = workspace_path / f"{cfg.memory_index}.json"
-
-        self.filename.touch(exist_ok=True)
-
-        file_content = b"{}"
-        with self.filename.open("w+b") as f:
-            f.write(file_content)
-
-        self.data = CacheContent()
-
-    def add(self, text: str):
-        """
-        Add text to our list of texts, add embedding as row to our
-            embeddings-matrix
-
-        Args:
-            text: str
-
-        Returns: None
-        """
-        if "Command Error:" in text:
-            return ""
-        self.data.texts.append(text)
-
-        embedding = get_ada_embedding(text)
-
-        vector = np.array(embedding).astype(np.float32)
-        vector = vector[np.newaxis, :]
-        self.data.embeddings = np.concatenate(
-            [
-                self.data.embeddings,
-                vector,
-            ],
-            axis=0,
-        )
-
-        with open(self.filename, "wb") as f:
-            out = orjson.dumps(self.data, option=SAVE_OPTIONS)
-            f.write(out)
-        return text
-
-    def clear(self) -> str:
-        """
-        Clears the data in memory.
-
-        Returns: A message indicating that the memory has been cleared.
-        """
-        self.data = CacheContent()
-        return "Obliviated"
-
-    def get(self, data: str) -> list[Any] | None:
-        """
-        Gets the data from the memory that is most relevant to the given data.
-
-        Args:
-            data: The data to compare to.
-
-        Returns: The most relevant data.
-        """
-        return self.get_relevant(data, 1)
-
-    def get_relevant(self, text: str, k: int) -> list[Any]:
-        """ "
-        matrix-vector mult to find score-for-each-row-of-matrix
-         get indices for top-k winning scores
-         return texts for those indices
-        Args:
-            text: str
-            k: int
-
-        Returns: List[str]
-        """
-        embedding = get_ada_embedding(text)
-
-        scores = np.dot(self.data.embeddings, embedding)
-
-        top_k_indices = np.argsort(scores)[-k:][::-1]
-
-        return [self.data.texts[i] for i in top_k_indices]
-
-    def get_stats(self) -> tuple[int, tuple[int, ...]]:
-        """
-        Returns: The stats of the local cache.
-        """
-        return len(self.data.texts), self.data.embeddings.shape

autogpt/memory/message_history.py

@@ -0,0 +1,241 @@
+from __future__ import annotations
+
+import copy
+import json
+from dataclasses import dataclass, field
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from autogpt.agent import Agent
+
+from autogpt.config import Config
+from autogpt.json_utils.utilities import (
+    LLM_DEFAULT_RESPONSE_FORMAT,
+    extract_json_from_response,
+    is_string_valid_json,
+)
+from autogpt.llm.base import ChatSequence, Message, MessageRole, MessageType
+from autogpt.llm.providers.openai import OPEN_AI_CHAT_MODELS
+from autogpt.llm.utils import count_string_tokens, create_chat_completion
+from autogpt.log_cycle.log_cycle import PROMPT_SUMMARY_FILE_NAME, SUMMARY_FILE_NAME
+from autogpt.logs import logger
+
+
+@dataclass
+class MessageHistory:
+    agent: Agent
+
+    messages: list[Message] = field(default_factory=list)
+    summary: str = "I was created"
+
+    last_trimmed_index: int = 0
+
+    def __getitem__(self, i: int):
+        return self.messages[i]
+
+    def __iter__(self):
+        return iter(self.messages)
+
+    def __len__(self):
+        return len(self.messages)
+
+    def add(
+        self,
+        role: MessageRole,
+        content: str,
+        type: MessageType | None = None,
+    ):
+        return self.append(Message(role, content, type))
+
+    def append(self, message: Message):
+        return self.messages.append(message)
+
+    def trim_messages(
+        self,
+        current_message_chain: list[Message],
+    ) -> tuple[Message, list[Message]]:
+        """
+        Returns a list of trimmed messages: messages which are in the message history
+        but not in current_message_chain.
+
+        Args:
+            current_message_chain (list[Message]): The messages currently in the context.
+
+        Returns:
+            Message: A message with the new running summary after adding the trimmed messages.
+            list[Message]: A list of messages that are in full_message_history with an index higher than last_trimmed_index and absent from current_message_chain.
+        """
+        # Select messages in full_message_history with an index higher than last_trimmed_index
+        new_messages = [
+            msg for i, msg in enumerate(self) if i > self.last_trimmed_index
+        ]
+
+        # Remove messages that are already present in current_message_chain
+        new_messages_not_in_chain = [
+            msg for msg in new_messages if msg not in current_message_chain
+        ]
+
+        if not new_messages_not_in_chain:
+            return self.summary_message(), []
+
+        new_summary_message = self.update_running_summary(
+            new_events=new_messages_not_in_chain
+        )
+
+        # Find the index of the last message processed
+        last_message = new_messages_not_in_chain[-1]
+        self.last_trimmed_index = self.messages.index(last_message)
+
+        return new_summary_message, new_messages_not_in_chain
+
+    def per_cycle(self, messages: list[Message] | None = None):
+        """
+        Yields:
+            Message: a message containing user input
+            Message: a message from the AI containing a proposed action
+            Message: the message containing the result of the AI's proposed action
+        """
+        messages = messages or self.messages
+        for i in range(0, len(messages) - 1):
+            ai_message = messages[i]
+            if ai_message.type != "ai_response":
+                continue
+            user_message = (
+                messages[i - 1] if i > 0 and messages[i - 1].role == "user" else None
+            )
+            result_message = messages[i + 1]
+            try:
+                assert is_string_valid_json(
+                    ai_message.content, LLM_DEFAULT_RESPONSE_FORMAT
+                ), "AI response is not a valid JSON object"
+                assert result_message.type == "action_result"
+
+                yield user_message, ai_message, result_message
+            except AssertionError as err:
+                logger.debug(
+                    f"Invalid item in message history: {err}; Messages: {messages[i-1:i+2]}"
+                )
+
+    def summary_message(self) -> Message:
+        return Message(
+            "system",
+            f"This reminds you of these events from your past: \n{self.summary}",
+        )
+
+    def update_running_summary(self, new_events: list[Message]) -> Message:
+        """
+        This function takes a list of dictionaries representing new events and combines them with the current summary,
+        focusing on key and potentially important information to remember. The updated summary is returned in a message
+        formatted in the 1st person past tense.
+
+        Args:
+            new_events (List[Dict]): A list of dictionaries containing the latest events to be added to the summary.
+
+        Returns:
+            str: A message containing the updated summary of actions, formatted in the 1st person past tense.
+
+        Example:
+            new_events = [{"event": "entered the kitchen."}, {"event": "found a scrawled note with the number 7"}]
+            update_running_summary(new_events)
+            # Returns: "This reminds you of these events from your past: \nI entered the kitchen and found a scrawled note saying 7."
+        """
+        cfg = Config()
+
+        if not new_events:
+            return self.summary_message()
+
+        # Create a copy of the new_events list to prevent modifying the original list
+        new_events = copy.deepcopy(new_events)
+
+        # Replace "assistant" with "you". This produces much better first person past tense results.
+        for event in new_events:
+            if event.role.lower() == "assistant":
+                event.role = "you"
+
+                # Remove "thoughts" dictionary from "content"
+                try:
+                    content_dict = extract_json_from_response(event.content)
+                    if "thoughts" in content_dict:
+                        del content_dict["thoughts"]
+                    event.content = json.dumps(content_dict)
+                except json.JSONDecodeError as e:
+                    logger.error(f"Error: Invalid JSON: {e}")
+                    if cfg.debug_mode:
+                        logger.error(f"{event.content}")
+
+            elif event.role.lower() == "system":
+                event.role = "your computer"
+
+            # Delete all user messages
+            elif event.role == "user":
+                new_events.remove(event)
+
+        # Summarize events and current summary in batch to a new running summary
+
+        # Assume an upper bound length for the summary prompt template, i.e. Your task is to create a concise running summary...., in summarize_batch func
+        # TODO make this default dynamic
+        prompt_template_length = 100
+        max_tokens = OPEN_AI_CHAT_MODELS.get(cfg.fast_llm_model).max_tokens
+        summary_tlength = count_string_tokens(str(self.summary), cfg.fast_llm_model)
+        batch = []
+        batch_tlength = 0
+
+        # TODO Can put a cap on length of total new events and drop some previous events to save API cost, but need to think thru more how to do it without losing the context
+        for event in new_events:
+            event_tlength = count_string_tokens(str(event), cfg.fast_llm_model)
+
+            if (
+                batch_tlength + event_tlength
+                > max_tokens - prompt_template_length - summary_tlength
+            ):
+                # The batch is full. Summarize it and start a new one.
+                self.summarize_batch(batch, cfg)
+                summary_tlength = count_string_tokens(
+                    str(self.summary), cfg.fast_llm_model
+                )
+                batch = [event]
+                batch_tlength = event_tlength
+            else:
+                batch.append(event)
+                batch_tlength += event_tlength
+
+        if batch:
+            # There's an unprocessed batch. Summarize it.
+            self.summarize_batch(batch, cfg)
+
+        return self.summary_message()
+
+    def summarize_batch(self, new_events_batch, cfg):
+        prompt = f'''Your task is to create a concise running summary of actions and information results in the provided text, focusing on key and potentially important information to remember.
+
+You will receive the current summary and your latest actions. Combine them, adding relevant key information from the latest development in 1st person past tense and keeping the summary concise.
+
+Summary So Far:
+"""
+{self.summary}
+"""
+
+Latest Development:
+"""
+{new_events_batch or "Nothing new happened."}
+"""
+'''
+
+        prompt = ChatSequence.for_model(cfg.fast_llm_model, [Message("user", prompt)])
+        self.agent.log_cycle_handler.log_cycle(
+            self.agent.ai_name,
+            self.agent.created_at,
+            self.agent.cycle_count,
+            prompt.raw(),
+            PROMPT_SUMMARY_FILE_NAME,
+        )
+
+        self.summary = create_chat_completion(prompt)
+
+        self.agent.log_cycle_handler.log_cycle(
+            self.agent.ai_name,
+            self.agent.created_at,
+            self.agent.cycle_count,
+            self.summary,
+            SUMMARY_FILE_NAME,
+        )

autogpt/memory/milvus.py

@@ -1,162 +0,0 @@
-""" Milvus memory storage provider."""
-import re
-
-from pymilvus import Collection, CollectionSchema, DataType, FieldSchema, connections
-
-from autogpt.config import Config
-from autogpt.llm import get_ada_embedding
-from autogpt.memory.base import MemoryProviderSingleton
-
-
-class MilvusMemory(MemoryProviderSingleton):
-    """Milvus memory storage provider."""
-
-    def __init__(self, cfg: Config) -> None:
-        """Construct a milvus memory storage connection.
-
-        Args:
-            cfg (Config): Auto-GPT global config.
-        """
-        self.configure(cfg)
-
-        connect_kwargs = {}
-        if self.username:
-            connect_kwargs["user"] = self.username
-            connect_kwargs["password"] = self.password
-
-        connections.connect(
-            **connect_kwargs,
-            uri=self.uri or "",
-            address=self.address or "",
-            secure=self.secure,
-        )
-
-        self.init_collection()
-
-    def configure(self, cfg: Config) -> None:
-        # init with configuration.
-        self.uri = None
-        self.address = cfg.milvus_addr
-        self.secure = cfg.milvus_secure
-        self.username = cfg.milvus_username
-        self.password = cfg.milvus_password
-        self.collection_name = cfg.milvus_collection
-        # use HNSW by default.
-        self.index_params = {
-            "metric_type": "IP",
-            "index_type": "HNSW",
-            "params": {"M": 8, "efConstruction": 64},
-        }
-
-        if (self.username is None) != (self.password is None):
-            raise ValueError(
-                "Both username and password must be set to use authentication for Milvus"
-            )
-
-        # configured address may be a full URL.
-        if re.match(r"^(https?|tcp)://", self.address) is not None:
-            self.uri = self.address
-            self.address = None
-
-            if self.uri.startswith("https"):
-                self.secure = True
-
-            # Zilliz Cloud requires AutoIndex.
-            if re.match(r"^https://(.*)\.zillizcloud\.(com|cn)", self.uri) is not None:
-                self.index_params = {
-                    "metric_type": "IP",
-                    "index_type": "AUTOINDEX",
-                    "params": {},
-                }
-
-    def init_collection(self) -> None:
-        """Initialize collection in vector database."""
-        fields = [
-            FieldSchema(name="pk", dtype=DataType.INT64, is_primary=True, auto_id=True),
-            FieldSchema(name="embeddings", dtype=DataType.FLOAT_VECTOR, dim=1536),
-            FieldSchema(name="raw_text", dtype=DataType.VARCHAR, max_length=65535),
-        ]
-
-        # create collection if not exist and load it.
-        self.schema = CollectionSchema(fields, "auto-gpt memory storage")
-        self.collection = Collection(self.collection_name, self.schema)
-        # create index if not exist.
-        if not self.collection.has_index():
-            self.collection.release()
-            self.collection.create_index(
-                "embeddings",
-                self.index_params,
-                index_name="embeddings",
-            )
-        self.collection.load()
-
-    def add(self, data) -> str:
-        """Add an embedding of data into memory.
-
-        Args:
-            data (str): The raw text to construct embedding index.
-
-        Returns:
-            str: log.
-        """
-        embedding = get_ada_embedding(data)
-        result = self.collection.insert([[embedding], [data]])
-        _text = (
-            "Inserting data into memory at primary key: "
-            f"{result.primary_keys[0]}:\n data: {data}"
-        )
-        return _text
-
-    def get(self, data):
-        """Return the most relevant data in memory.
-        Args:
-            data: The data to compare to.
-        """
-        return self.get_relevant(data, 1)
-
-    def clear(self) -> str:
-        """Drop the index in memory.
-
-        Returns:
-            str: log.
-        """
-        self.collection.drop()
-        self.collection = Collection(self.collection_name, self.schema)
-        self.collection.create_index(
-            "embeddings",
-            self.index_params,
-            index_name="embeddings",
-        )
-        self.collection.load()
-        return "Obliviated"
-
-    def get_relevant(self, data: str, num_relevant: int = 5):
-        """Return the top-k relevant data in memory.
-        Args:
-            data: The data to compare to.
-            num_relevant (int, optional): The max number of relevant data.
-                Defaults to 5.
-
-        Returns:
-            list: The top-k relevant data.
-        """
-        # search the embedding and return the most relevant text.
-        embedding = get_ada_embedding(data)
-        search_params = {
-            "metrics_type": "IP",
-            "params": {"nprobe": 8},
-        }
-        result = self.collection.search(
-            [embedding],
-            "embeddings",
-            search_params,
-            num_relevant,
-            output_fields=["raw_text"],
-        )
-        return [item.entity.value_of_field("raw_text") for item in result[0]]
-
-    def get_stats(self) -> str:
-        """
-        Returns: The stats of the milvus cache.
-        """
-        return f"Entities num: {self.collection.num_entities}"

autogpt/memory/no_memory.py

@@ -1,73 +0,0 @@
-"""A class that does not store any data. This is the default memory provider."""
-from __future__ import annotations
-
-from typing import Any
-
-from autogpt.memory.base import MemoryProviderSingleton
-
-
-class NoMemory(MemoryProviderSingleton):
-    """
-    A class that does not store any data. This is the default memory provider.
-    """
-
-    def __init__(self, cfg):
-        """
-        Initializes the NoMemory provider.
-
-        Args:
-            cfg: The config object.
-
-        Returns: None
-        """
-        pass
-
-    def add(self, data: str) -> str:
-        """
-        Adds a data point to the memory. No action is taken in NoMemory.
-
-        Args:
-            data: The data to add.
-
-        Returns: An empty string.
-        """
-        return ""
-
-    def get(self, data: str) -> list[Any] | None:
-        """
-        Gets the data from the memory that is most relevant to the given data.
-        NoMemory always returns None.
-
-        Args:
-            data: The data to compare to.
-
-        Returns: None
-        """
-        return None
-
-    def clear(self) -> str:
-        """
-        Clears the memory. No action is taken in NoMemory.
-
-        Returns: An empty string.
-        """
-        return ""
-
-    def get_relevant(self, data: str, num_relevant: int = 5) -> list[Any] | None:
-        """
-        Returns all the data in the memory that is relevant to the given data.
-        NoMemory always returns None.
-
-        Args:
-            data: The data to compare to.
-            num_relevant: The number of relevant data to return.
-
-        Returns: None
-        """
-        return None
-
-    def get_stats(self):
-        """
-        Returns: An empty dictionary as there are no stats in NoMemory.
-        """
-        return {}

autogpt/memory/pinecone.py

@@ -1,75 +0,0 @@
-import pinecone
-from colorama import Fore, Style
-
-from autogpt.llm import get_ada_embedding
-from autogpt.logs import logger
-from autogpt.memory.base import MemoryProviderSingleton
-
-
-class PineconeMemory(MemoryProviderSingleton):
-    def __init__(self, cfg):
-        pinecone_api_key = cfg.pinecone_api_key
-        pinecone_region = cfg.pinecone_region
-        pinecone.init(api_key=pinecone_api_key, environment=pinecone_region)
-        dimension = 1536
-        metric = "cosine"
-        pod_type = "p1"
-        table_name = "auto-gpt"
-        # this assumes we don't start with memory.
-        # for now this works.
-        # we'll need a more complicated and robust system if we want to start with
-        #  memory.
-        self.vec_num = 0
-
-        try:
-            pinecone.whoami()
-        except Exception as e:
-            logger.typewriter_log(
-                "FAILED TO CONNECT TO PINECONE",
-                Fore.RED,
-                Style.BRIGHT + str(e) + Style.RESET_ALL,
-            )
-            logger.double_check(
-                "Please ensure you have setup and configured Pinecone properly for use."
-                + f"You can check out {Fore.CYAN + Style.BRIGHT}"
-                "https://github.com/Torantulino/Auto-GPT#-pinecone-api-key-setup"
-                f"{Style.RESET_ALL} to ensure you've set up everything correctly."
-            )
-            exit(1)
-
-        if table_name not in pinecone.list_indexes():
-            pinecone.create_index(
-                table_name, dimension=dimension, metric=metric, pod_type=pod_type
-            )
-        self.index = pinecone.Index(table_name)
-
-    def add(self, data):
-        vector = get_ada_embedding(data)
-        # no metadata here. We may wish to change that long term.
-        self.index.upsert([(str(self.vec_num), vector, {"raw_text": data})])
-        _text = f"Inserting data into memory at index: {self.vec_num}:\n data: {data}"
-        self.vec_num += 1
-        return _text
-
-    def get(self, data):
-        return self.get_relevant(data, 1)
-
-    def clear(self):
-        self.index.delete(deleteAll=True)
-        return "Obliviated"
-
-    def get_relevant(self, data, num_relevant=5):
-        """
-        Returns all the data in the memory that is relevant to the given data.
-        :param data: The data to compare to.
-        :param num_relevant: The number of relevant data to return. Defaults to 5
-        """
-        query_embedding = get_ada_embedding(data)
-        results = self.index.query(
-            query_embedding, top_k=num_relevant, include_metadata=True
-        )
-        sorted_results = sorted(results.matches, key=lambda x: x.score)
-        return [str(item["metadata"]["raw_text"]) for item in sorted_results]
-
-    def get_stats(self):
-        return self.index.describe_index_stats()

autogpt/memory/redismem.py

@@ -1,156 +0,0 @@
-"""Redis memory provider."""
-from __future__ import annotations
-
-from typing import Any
-
-import numpy as np
-import redis
-from colorama import Fore, Style
-from redis.commands.search.field import TextField, VectorField
-from redis.commands.search.indexDefinition import IndexDefinition, IndexType
-from redis.commands.search.query import Query
-
-from autogpt.llm import get_ada_embedding
-from autogpt.logs import logger
-from autogpt.memory.base import MemoryProviderSingleton
-
-SCHEMA = [
-    TextField("data"),
-    VectorField(
-        "embedding",
-        "HNSW",
-        {"TYPE": "FLOAT32", "DIM": 1536, "DISTANCE_METRIC": "COSINE"},
-    ),
-]
-
-
-class RedisMemory(MemoryProviderSingleton):
-    def __init__(self, cfg):
-        """
-        Initializes the Redis memory provider.
-
-        Args:
-            cfg: The config object.
-
-        Returns: None
-        """
-        redis_host = cfg.redis_host
-        redis_port = cfg.redis_port
-        redis_password = cfg.redis_password
-        self.dimension = 1536
-        self.redis = redis.Redis(
-            host=redis_host,
-            port=redis_port,
-            password=redis_password,
-            db=0,  # Cannot be changed
-        )
-        self.cfg = cfg
-
-        # Check redis connection
-        try:
-            self.redis.ping()
-        except redis.ConnectionError as e:
-            logger.typewriter_log(
-                "FAILED TO CONNECT TO REDIS",
-                Fore.RED,
-                Style.BRIGHT + str(e) + Style.RESET_ALL,
-            )
-            logger.double_check(
-                "Please ensure you have setup and configured Redis properly for use. "
-                + f"You can check out {Fore.CYAN + Style.BRIGHT}"
-                f"https://github.com/Torantulino/Auto-GPT#redis-setup{Style.RESET_ALL}"
-                " to ensure you've set up everything correctly."
-            )
-            exit(1)
-
-        if cfg.wipe_redis_on_start:
-            self.redis.flushall()
-        try:
-            self.redis.ft(f"{cfg.memory_index}").create_index(
-                fields=SCHEMA,
-                definition=IndexDefinition(
-                    prefix=[f"{cfg.memory_index}:"], index_type=IndexType.HASH
-                ),
-            )
-        except Exception as e:
-            logger.warn("Error creating Redis search index: ", e)
-        existing_vec_num = self.redis.get(f"{cfg.memory_index}-vec_num")
-        self.vec_num = int(existing_vec_num.decode("utf-8")) if existing_vec_num else 0
-
-    def add(self, data: str) -> str:
-        """
-        Adds a data point to the memory.
-
-        Args:
-            data: The data to add.
-
-        Returns: Message indicating that the data has been added.
-        """
-        if "Command Error:" in data:
-            return ""
-        vector = get_ada_embedding(data)
-        vector = np.array(vector).astype(np.float32).tobytes()
-        data_dict = {b"data": data, "embedding": vector}
-        pipe = self.redis.pipeline()
-        pipe.hset(f"{self.cfg.memory_index}:{self.vec_num}", mapping=data_dict)
-        _text = (
-            f"Inserting data into memory at index: {self.vec_num}:\n" f"data: {data}"
-        )
-        self.vec_num += 1
-        pipe.set(f"{self.cfg.memory_index}-vec_num", self.vec_num)
-        pipe.execute()
-        return _text
-
-    def get(self, data: str) -> list[Any] | None:
-        """
-        Gets the data from the memory that is most relevant to the given data.
-
-        Args:
-            data: The data to compare to.
-
-        Returns: The most relevant data.
-        """
-        return self.get_relevant(data, 1)
-
-    def clear(self) -> str:
-        """
-        Clears the redis server.
-
-        Returns: A message indicating that the memory has been cleared.
-        """
-        self.redis.flushall()
-        return "Obliviated"
-
-    def get_relevant(self, data: str, num_relevant: int = 5) -> list[Any] | None:
-        """
-        Returns all the data in the memory that is relevant to the given data.
-        Args:
-            data: The data to compare to.
-            num_relevant: The number of relevant data to return.
-
-        Returns: A list of the most relevant data.
-        """
-        query_embedding = get_ada_embedding(data)
-        base_query = f"*=>[KNN {num_relevant} @embedding $vector AS vector_score]"
-        query = (
-            Query(base_query)
-            .return_fields("data", "vector_score")
-            .sort_by("vector_score")
-            .dialect(2)
-        )
-        query_vector = np.array(query_embedding).astype(np.float32).tobytes()
-
-        try:
-            results = self.redis.ft(f"{self.cfg.memory_index}").search(
-                query, query_params={"vector": query_vector}
-            )
-        except Exception as e:
-            logger.warn("Error calling Redis search: ", e)
-            return None
-        return [result.data for result in results.docs]
-
-    def get_stats(self):
-        """
-        Returns: The stats of the memory index.
-        """
-        return self.redis.ft(f"{self.cfg.memory_index}").info()

autogpt/memory/vector/__init__.py

@@ -0,0 +1,138 @@
+from autogpt.config import Config
+from autogpt.logs import logger
+
+from .memory_item import MemoryItem, MemoryItemRelevance
+from .providers.base import VectorMemoryProvider as VectorMemory
+from .providers.json_file import JSONFileMemory
+from .providers.no_memory import NoMemory
+
+# List of supported memory backends
+# Add a backend to this list if the import attempt is successful
+supported_memory = ["json_file", "no_memory"]
+
+# try:
+#     from .providers.redis import RedisMemory
+
+#     supported_memory.append("redis")
+# except ImportError:
+#     RedisMemory = None
+
+# try:
+#     from .providers.pinecone import PineconeMemory
+
+#     supported_memory.append("pinecone")
+# except ImportError:
+#     PineconeMemory = None
+
+# try:
+#     from .providers.weaviate import WeaviateMemory
+
+#     supported_memory.append("weaviate")
+# except ImportError:
+#     WeaviateMemory = None
+
+# try:
+#     from .providers.milvus import MilvusMemory
+
+#     supported_memory.append("milvus")
+# except ImportError:
+#     MilvusMemory = None
+
+
+def get_memory(cfg: Config, init=False) -> VectorMemory:
+    memory = None
+
+    match cfg.memory_backend:
+        case "json_file":
+            memory = JSONFileMemory(cfg)
+
+        case "pinecone":
+            raise NotImplementedError(
+                "The Pinecone memory backend has been rendered incompatible by work on "
+                "the memory system, and was removed. Whether support will be added back "
+                "in the future is subject to discussion, feel free to pitch in: "
+                "https://github.com/Significant-Gravitas/Auto-GPT/discussions/4280"
+            )
+            # if not PineconeMemory:
+            #     logger.warn(
+            #         "Error: Pinecone is not installed. Please install pinecone"
+            #         " to use Pinecone as a memory backend."
+            #     )
+            # else:
+            #     memory = PineconeMemory(cfg)
+            #     if init:
+            #         memory.clear()
+
+        case "redis":
+            raise NotImplementedError(
+                "The Redis memory backend has been rendered incompatible by work on "
+                "the memory system, and has been removed temporarily."
+            )
+            # if not RedisMemory:
+            #     logger.warn(
+            #         "Error: Redis is not installed. Please install redis-py to"
+            #         " use Redis as a memory backend."
+            #     )
+            # else:
+            #     memory = RedisMemory(cfg)
+
+        case "weaviate":
+            raise NotImplementedError(
+                "The Weaviate memory backend has been rendered incompatible by work on "
+                "the memory system, and was removed. Whether support will be added back "
+                "in the future is subject to discussion, feel free to pitch in: "
+                "https://github.com/Significant-Gravitas/Auto-GPT/discussions/4280"
+            )
+            # if not WeaviateMemory:
+            #     logger.warn(
+            #         "Error: Weaviate is not installed. Please install weaviate-client to"
+            #         " use Weaviate as a memory backend."
+            #     )
+            # else:
+            #     memory = WeaviateMemory(cfg)
+
+        case "milvus":
+            raise NotImplementedError(
+                "The Milvus memory backend has been rendered incompatible by work on "
+                "the memory system, and was removed. Whether support will be added back "
+                "in the future is subject to discussion, feel free to pitch in: "
+                "https://github.com/Significant-Gravitas/Auto-GPT/discussions/4280"
+            )
+            # if not MilvusMemory:
+            #     logger.warn(
+            #         "Error: pymilvus sdk is not installed."
+            #         "Please install pymilvus to use Milvus or Zilliz Cloud as memory backend."
+            #     )
+            # else:
+            #     memory = MilvusMemory(cfg)
+
+        case "no_memory":
+            memory = NoMemory()
+
+        case _:
+            raise ValueError(
+                f"Unknown memory backend '{cfg.memory_backend}'. Please check your config."
+            )
+
+    if memory is None:
+        memory = JSONFileMemory(cfg)
+
+    return memory
+
+
+def get_supported_memory_backends():
+    return supported_memory
+
+
+__all__ = [
+    "get_memory",
+    "MemoryItem",
+    "MemoryItemRelevance",
+    "JSONFileMemory",
+    "NoMemory",
+    "VectorMemory",
+    # "RedisMemory",
+    # "PineconeMemory",
+    # "MilvusMemory",
+    # "WeaviateMemory",
+]

autogpt/memory/vector/memory_item.py

@@ -0,0 +1,223 @@
+from __future__ import annotations
+
+import dataclasses
+import json
+from typing import Literal
+
+import numpy as np
+
+from autogpt.config import Config
+from autogpt.llm import Message
+from autogpt.llm.utils import count_string_tokens
+from autogpt.logs import logger
+from autogpt.processing.text import chunk_content, split_text, summarize_text
+
+from .utils import Embedding, get_embedding
+
+MemoryDocType = Literal["webpage", "text_file", "code_file", "agent_history"]
+
+
+@dataclasses.dataclass
+class MemoryItem:
+    """Memory object containing raw content as well as embeddings"""
+
+    raw_content: str
+    summary: str
+    chunks: list[str]
+    chunk_summaries: list[str]
+    e_summary: Embedding
+    e_chunks: list[Embedding]
+    metadata: dict
+
+    def relevance_for(self, query: str, e_query: Embedding | None = None):
+        return MemoryItemRelevance.of(self, query, e_query)
+
+    @staticmethod
+    def from_text(
+        text: str,
+        source_type: MemoryDocType,
+        metadata: dict = {},
+        how_to_summarize: str | None = None,
+        question_for_summary: str | None = None,
+    ):
+        cfg = Config()
+        logger.debug(f"Memorizing text:\n{'-'*32}\n{text}\n{'-'*32}\n")
+
+        chunks = [
+            chunk
+            for chunk, _ in (
+                split_text(text, cfg.embedding_model)
+                if source_type != "code_file"
+                else chunk_content(text, cfg.embedding_model)
+            )
+        ]
+        logger.debug("Chunks: " + str(chunks))
+
+        chunk_summaries = [
+            summary
+            for summary, _ in [
+                summarize_text(
+                    text_chunk,
+                    instruction=how_to_summarize,
+                    question=question_for_summary,
+                )
+                for text_chunk in chunks
+            ]
+        ]
+        logger.debug("Chunk summaries: " + str(chunk_summaries))
+
+        e_chunks = get_embedding(chunks)
+
+        summary = (
+            chunk_summaries[0]
+            if len(chunks) == 1
+            else summarize_text(
+                "\n\n".join(chunk_summaries),
+                instruction=how_to_summarize,
+                question=question_for_summary,
+            )[0]
+        )
+        logger.debug("Total summary: " + summary)
+
+        # TODO: investigate search performance of weighted average vs summary
+        # e_average = np.average(e_chunks, axis=0, weights=[len(c) for c in chunks])
+        e_summary = get_embedding(summary)
+
+        metadata["source_type"] = source_type
+
+        return MemoryItem(
+            text,
+            summary,
+            chunks,
+            chunk_summaries,
+            e_summary,
+            e_chunks,
+            metadata=metadata,
+        )
+
+    @staticmethod
+    def from_text_file(content: str, path: str):
+        return MemoryItem.from_text(content, "text_file", {"location": path})
+
+    @staticmethod
+    def from_code_file(content: str, path: str):
+        # TODO: implement tailored code memories
+        return MemoryItem.from_text(content, "code_file", {"location": path})
+
+    @staticmethod
+    def from_ai_action(ai_message: Message, result_message: Message):
+        # The result_message contains either user feedback
+        # or the result of the command specified in ai_message
+
+        if ai_message["role"] != "assistant":
+            raise ValueError(f"Invalid role on 'ai_message': {ai_message['role']}")
+
+        result = (
+            result_message["content"]
+            if result_message["content"].startswith("Command")
+            else "None"
+        )
+        user_input = (
+            result_message["content"]
+            if result_message["content"].startswith("Human feedback")
+            else "None"
+        )
+        memory_content = (
+            f"Assistant Reply: {ai_message['content']}"
+            "\n\n"
+            f"Result: {result}"
+            "\n\n"
+            f"Human Feedback: {user_input}"
+        )
+
+        return MemoryItem.from_text(
+            text=memory_content,
+            source_type="agent_history",
+            how_to_summarize="if possible, also make clear the link between the command in the assistant's response and the command result. Do not mention the human feedback if there is none",
+        )
+
+    @staticmethod
+    def from_webpage(content: str, url: str, question: str | None = None):
+        return MemoryItem.from_text(
+            text=content,
+            source_type="webpage",
+            metadata={"location": url},
+            question_for_summary=question,
+        )
+
+    def dump(self) -> str:
+        token_length = count_string_tokens(self.raw_content, Config().embedding_model)
+        return f"""
+=============== MemoryItem ===============
+Length: {token_length} tokens in {len(self.e_chunks)} chunks
+Metadata: {json.dumps(self.metadata, indent=2)}
+---------------- SUMMARY -----------------
+{self.summary}
+------------------ RAW -------------------
+{self.raw_content}
+==========================================
+"""
+
+
+@dataclasses.dataclass
+class MemoryItemRelevance:
+    """
+    Class that encapsulates memory relevance search functionality and data.
+    Instances contain a MemoryItem and its relevance scores for a given query.
+    """
+
+    memory_item: MemoryItem
+    for_query: str
+    summary_relevance_score: float
+    chunk_relevance_scores: list[float]
+
+    @staticmethod
+    def of(
+        memory_item: MemoryItem, for_query: str, e_query: Embedding | None = None
+    ) -> MemoryItemRelevance:
+        e_query = e_query or get_embedding(for_query)
+        _, srs, crs = MemoryItemRelevance.calculate_scores(memory_item, e_query)
+        return MemoryItemRelevance(
+            for_query=for_query,
+            memory_item=memory_item,
+            summary_relevance_score=srs,
+            chunk_relevance_scores=crs,
+        )
+
+    @staticmethod
+    def calculate_scores(
+        memory: MemoryItem, compare_to: Embedding
+    ) -> tuple[float, float, list[float]]:
+        """
+        Calculates similarity between given embedding and all embeddings of the memory
+
+        Returns:
+            float: the aggregate (max) relevance score of the memory
+            float: the relevance score of the memory summary
+            list: the relevance scores of the memory chunks
+        """
+        summary_relevance_score = np.dot(memory.e_summary, compare_to)
+        chunk_relevance_scores = np.dot(memory.e_chunks, compare_to)
+        logger.debug(f"Relevance of summary: {summary_relevance_score}")
+        logger.debug(f"Relevance of chunks: {chunk_relevance_scores}")
+
+        relevance_scores = [summary_relevance_score, *chunk_relevance_scores]
+        logger.debug(f"Relevance scores: {relevance_scores}")
+        return max(relevance_scores), summary_relevance_score, chunk_relevance_scores
+
+    @property
+    def score(self) -> float:
+        """The aggregate relevance score of the memory item for the given query"""
+        return max([self.summary_relevance_score, *self.chunk_relevance_scores])
+
+    @property
+    def most_relevant_chunk(self) -> tuple[str, float]:
+        """The most relevant chunk of the memory item + its score for the given query"""
+        i_relmax = np.argmax(self.chunk_relevance_scores)
+        return self.memory_item.chunks[i_relmax], self.chunk_relevance_scores[i_relmax]
+
+    def __str__(self):
+        return (
+            f"{self.memory_item.summary} ({self.summary_relevance_score}) "
+            f"{self.chunk_relevance_scores}"
+        )

autogpt/memory/vector/providers/__init__.py

@@ -0,0 +1,7 @@
+from .json_file import JSONFileMemory
+from .no_memory import NoMemory
+
+__all__ = [
+    "JSONFileMemory",
+    "NoMemory",
+]

autogpt/memory/vector/providers/base.py

@@ -0,0 +1,74 @@
+import abc
+import functools
+from typing import MutableSet, Sequence
+
+import numpy as np
+
+from autogpt.config.config import Config
+from autogpt.logs import logger
+from autogpt.singleton import AbstractSingleton
+
+from .. import MemoryItem, MemoryItemRelevance
+from ..utils import Embedding, get_embedding
+
+
+class VectorMemoryProvider(MutableSet[MemoryItem], AbstractSingleton):
+    @abc.abstractmethod
+    def __init__(self, config: Config):
+        pass
+
+    def get(self, query: str) -> MemoryItemRelevance | None:
+        """
+        Gets the data from the memory that is most relevant to the given query.
+
+        Args:
+            data: The data to compare to.
+
+        Returns: The most relevant Memory
+        """
+        result = self.get_relevant(query, 1)
+        return result[0] if result else None
+
+    def get_relevant(self, query: str, k: int) -> Sequence[MemoryItemRelevance]:
+        """
+        Returns the top-k most relevant memories for the given query
+
+        Args:
+            query: the query to compare stored memories to
+            k: the number of relevant memories to fetch
+
+        Returns:
+            list[MemoryItemRelevance] containing the top [k] relevant memories
+        """
+        if len(self) < 1:
+            return []
+
+        logger.debug(
+            f"Searching for {k} relevant memories for query '{query}'; "
+            f"{len(self)} memories in index"
+        )
+
+        relevances = self.score_memories_for_relevance(query)
+        logger.debug(f"Memory relevance scores: {[str(r) for r in relevances]}")
+
+        # take last k items and reverse
+        top_k_indices = np.argsort([r.score for r in relevances])[-k:][::-1]
+
+        return [relevances[i] for i in top_k_indices]
+
+    def score_memories_for_relevance(
+        self, for_query: str
+    ) -> Sequence[MemoryItemRelevance]:
+        """
+        Returns MemoryItemRelevance for every memory in the index.
+        Implementations may override this function for performance purposes.
+        """
+        e_query: Embedding = get_embedding(for_query)
+        return [m.relevance_for(for_query, e_query) for m in self]
+
+    def get_stats(self) -> tuple[int, int]:
+        """
+        Returns:
+            tuple (n_memories: int, n_chunks: int): the stats of the memory index
+        """
+        return len(self), functools.reduce(lambda t, m: t + len(m.e_chunks), self, 0)

autogpt/memory/vector/providers/json_file.py

@@ -0,0 +1,68 @@
+from __future__ import annotations
+
+from pathlib import Path
+from typing import Iterator
+
+import orjson
+
+from autogpt.config import Config
+from autogpt.logs import logger
+
+from ..memory_item import MemoryItem
+from .base import VectorMemoryProvider
+
+
+class JSONFileMemory(VectorMemoryProvider):
+    """Memory backend that stores memories in a JSON file"""
+
+    SAVE_OPTIONS = orjson.OPT_SERIALIZE_NUMPY | orjson.OPT_SERIALIZE_DATACLASS
+
+    file_path: Path
+    memories: list[MemoryItem]
+
+    def __init__(self, cfg: Config) -> None:
+        """Initialize a class instance
+
+        Args:
+            cfg: Config object
+
+        Returns:
+            None
+        """
+        workspace_path = Path(cfg.workspace_path)
+        self.file_path = workspace_path / f"{cfg.memory_index}.json"
+        self.file_path.touch()
+        logger.debug(f"Initialized {__name__} with index path {self.file_path}")
+
+        self.memories = []
+        self.save_index()
+
+    def __iter__(self) -> Iterator[MemoryItem]:
+        return iter(self.memories)
+
+    def __contains__(self, x: MemoryItem) -> bool:
+        return x in self.memories
+
+    def __len__(self) -> int:
+        return len(self.memories)
+
+    def add(self, item: MemoryItem):
+        self.memories.append(item)
+        self.save_index()
+        return len(self.memories)
+
+    def discard(self, item: MemoryItem):
+        try:
+            self.remove(item)
+        except:
+            pass
+
+    def clear(self):
+        """Clears the data in memory."""
+        self.memories.clear()
+        self.save_index()
+
+    def save_index(self):
+        logger.debug(f"Saving memory index to file {self.file_path}")
+        with self.file_path.open("wb") as f:
+            return f.write(orjson.dumps(self.memories, option=self.SAVE_OPTIONS))

autogpt/memory/vector/providers/no_memory.py

@@ -0,0 +1,36 @@
+"""A class that does not store any data. This is the default memory provider."""
+from __future__ import annotations
+
+from typing import Iterator, Optional
+
+from autogpt.config.config import Config
+
+from .. import MemoryItem
+from .base import VectorMemoryProvider
+
+
+class NoMemory(VectorMemoryProvider):
+    """
+    A class that does not store any data. This is the default memory provider.
+    """
+
+    def __init__(self, config: Optional[Config] = None):
+        pass
+
+    def __iter__(self) -> Iterator[MemoryItem]:
+        return iter([])
+
+    def __contains__(self, x: MemoryItem) -> bool:
+        return False
+
+    def __len__(self) -> int:
+        return 0
+
+    def add(self, item: MemoryItem):
+        pass
+
+    def discard(self, item: MemoryItem):
+        pass
+
+    def clear(self):
+        pass

autogpt/memory/vector/utils.py

@@ -0,0 +1,70 @@
+from typing import Any, overload
+
+import numpy as np
+import openai
+
+from autogpt.config import Config
+from autogpt.llm.utils import metered, retry_openai_api
+from autogpt.logs import logger
+
+Embedding = list[np.float32] | np.ndarray[Any, np.dtype[np.float32]]
+"""Embedding vector"""
+TText = list[int]
+"""Token array representing text"""
+
+
+@overload
+def get_embedding(input: str | TText) -> Embedding:
+    ...
+
+
+@overload
+def get_embedding(input: list[str] | list[TText]) -> list[Embedding]:
+    ...
+
+
+@metered
+@retry_openai_api()
+def get_embedding(
+    input: str | TText | list[str] | list[TText],
+) -> Embedding | list[Embedding]:
+    """Get an embedding from the ada model.
+
+    Args:
+        input: Input text to get embeddings for, encoded as a string or array of tokens.
+            Multiple inputs may be given as a list of strings or token arrays.
+
+    Returns:
+        List[float]: The embedding.
+    """
+    cfg = Config()
+    multiple = isinstance(input, list) and all(not isinstance(i, int) for i in input)
+
+    if isinstance(input, str):
+        input = input.replace("\n", " ")
+    elif multiple and isinstance(input[0], str):
+        input = [text.replace("\n", " ") for text in input]
+
+    model = cfg.embedding_model
+    if cfg.use_azure:
+        kwargs = {"engine": cfg.get_azure_deployment_id_for_model(model)}
+    else:
+        kwargs = {"model": model}
+
+    logger.debug(
+        f"Getting embedding{f's for {len(input)} inputs' if multiple else ''}"
+        f" with model '{model}'"
+        + (f" via Azure deployment '{kwargs['engine']}'" if cfg.use_azure else "")
+    )
+
+    embeddings = openai.Embedding.create(
+        input=input,
+        api_key=cfg.openai_api_key,
+        **kwargs,
+    ).data
+
+    if not multiple:
+        return embeddings[0]["embedding"]
+
+    embeddings = sorted(embeddings, key=lambda x: x["index"])
+    return [d["embedding"] for d in embeddings]

autogpt/memory/weaviate.py

@@ -1,127 +0,0 @@
-import weaviate
-from weaviate import Client
-from weaviate.embedded import EmbeddedOptions
-from weaviate.util import generate_uuid5
-
-from autogpt.llm import get_ada_embedding
-from autogpt.logs import logger
-from autogpt.memory.base import MemoryProviderSingleton
-
-
-def default_schema(weaviate_index):
-    return {
-        "class": weaviate_index,
-        "properties": [
-            {
-                "name": "raw_text",
-                "dataType": ["text"],
-                "description": "original text for the embedding",
-            }
-        ],
-    }
-
-
-class WeaviateMemory(MemoryProviderSingleton):
-    def __init__(self, cfg):
-        auth_credentials = self._build_auth_credentials(cfg)
-
-        url = f"{cfg.weaviate_protocol}://{cfg.weaviate_host}:{cfg.weaviate_port}"
-
-        if cfg.use_weaviate_embedded:
-            self.client = Client(
-                embedded_options=EmbeddedOptions(
-                    hostname=cfg.weaviate_host,
-                    port=int(cfg.weaviate_port),
-                    persistence_data_path=cfg.weaviate_embedded_path,
-                )
-            )
-
-            logger.info(
-                f"Weaviate Embedded running on: {url} with persistence path: {cfg.weaviate_embedded_path}"
-            )
-        else:
-            self.client = Client(url, auth_client_secret=auth_credentials)
-
-        self.index = WeaviateMemory.format_classname(cfg.memory_index)
-        self._create_schema()
-
-    @staticmethod
-    def format_classname(index):
-        # weaviate uses capitalised index names
-        # The python client uses the following code to format
-        # index names before the corresponding class is created
-        index = index.replace("-", "_")
-        if len(index) == 1:
-            return index.capitalize()
-        return index[0].capitalize() + index[1:]
-
-    def _create_schema(self):
-        schema = default_schema(self.index)
-        if not self.client.schema.contains(schema):
-            self.client.schema.create_class(schema)
-
-    def _build_auth_credentials(self, cfg):
-        if cfg.weaviate_username and cfg.weaviate_password:
-            return weaviate.AuthClientPassword(
-                cfg.weaviate_username, cfg.weaviate_password
-            )
-        if cfg.weaviate_api_key:
-            return weaviate.AuthApiKey(api_key=cfg.weaviate_api_key)
-        else:
-            return None
-
-    def add(self, data):
-        vector = get_ada_embedding(data)
-
-        doc_uuid = generate_uuid5(data, self.index)
-        data_object = {"raw_text": data}
-
-        with self.client.batch as batch:
-            batch.add_data_object(
-                uuid=doc_uuid,
-                data_object=data_object,
-                class_name=self.index,
-                vector=vector,
-            )
-
-        return f"Inserting data into memory at uuid: {doc_uuid}:\n data: {data}"
-
-    def get(self, data):
-        return self.get_relevant(data, 1)
-
-    def clear(self):
-        self.client.schema.delete_all()
-
-        # weaviate does not yet have a neat way to just remove the items in an index
-        # without removing the entire schema, therefore we need to re-create it
-        # after a call to delete_all
-        self._create_schema()
-
-        return "Obliterated"
-
-    def get_relevant(self, data, num_relevant=5):
-        query_embedding = get_ada_embedding(data)
-        try:
-            results = (
-                self.client.query.get(self.index, ["raw_text"])
-                .with_near_vector({"vector": query_embedding, "certainty": 0.7})
-                .with_limit(num_relevant)
-                .do()
-            )
-
-            if len(results["data"]["Get"][self.index]) > 0:
-                return [
-                    str(item["raw_text"]) for item in results["data"]["Get"][self.index]
-                ]
-            else:
-                return []
-
-        except Exception as err:
-            logger.warn(f"Unexpected error {err=}, {type(err)=}")
-            return []
-
-    def get_stats(self):
-        result = self.client.query.aggregate(self.index).with_meta_count().do()
-        class_data = result["data"]["Aggregate"][self.index]
-
-        return class_data[0]["meta"] if class_data else {}

autogpt/memory_management/store_memory.py

@@ -1,33 +0,0 @@
-from autogpt.json_utils.utilities import (
-    LLM_DEFAULT_RESPONSE_FORMAT,
-    is_string_valid_json,
-)
-from autogpt.logs import logger
-
-
-def format_memory(assistant_reply, next_message_content):
-    # the next_message_content is a variable to stores either the user_input or the command following the assistant_reply
-    result = (
-        "None" if next_message_content.startswith("Command") else next_message_content
-    )
-    user_input = (
-        "None"
-        if next_message_content.startswith("Human feedback")
-        else next_message_content
-    )
-
-    return f"Assistant Reply: {assistant_reply}\nResult: {result}\nHuman Feedback:{user_input}"
-
-
-def save_memory_trimmed_from_context_window(
-    full_message_history, next_message_to_add_index, permanent_memory
-):
-    while next_message_to_add_index >= 0:
-        message_content = full_message_history[next_message_to_add_index]["content"]
-        if is_string_valid_json(message_content, LLM_DEFAULT_RESPONSE_FORMAT):
-            next_message = full_message_history[next_message_to_add_index + 1]
-            memory_to_add = format_memory(message_content, next_message["content"])
-            logger.debug(f"Storing the following memory: {memory_to_add}")
-            permanent_memory.add(memory_to_add)
-
-        next_message_to_add_index -= 1

autogpt/memory_management/summary_memory.py

@@ -1,112 +0,0 @@
-import json
-from typing import Dict, List, Tuple
-
-from autogpt.config import Config
-from autogpt.llm.llm_utils import create_chat_completion
-
-cfg = Config()
-
-
-def get_newly_trimmed_messages(
-    full_message_history: List[Dict[str, str]],
-    current_context: List[Dict[str, str]],
-    last_memory_index: int,
-) -> Tuple[List[Dict[str, str]], int]:
-    """
-    This function returns a list of dictionaries contained in full_message_history
-    with an index higher than prev_index that are absent from current_context.
-
-    Args:
-        full_message_history (list): A list of dictionaries representing the full message history.
-        current_context (list): A list of dictionaries representing the current context.
-        last_memory_index (int): An integer representing the previous index.
-
-    Returns:
-        list: A list of dictionaries that are in full_message_history with an index higher than last_memory_index and absent from current_context.
-        int: The new index value for use in the next loop.
-    """
-    # Select messages in full_message_history with an index higher than last_memory_index
-    new_messages = [
-        msg for i, msg in enumerate(full_message_history) if i > last_memory_index
-    ]
-
-    # Remove messages that are already present in current_context
-    new_messages_not_in_context = [
-        msg for msg in new_messages if msg not in current_context
-    ]
-
-    # Find the index of the last message processed
-    new_index = last_memory_index
-    if new_messages_not_in_context:
-        last_message = new_messages_not_in_context[-1]
-        new_index = full_message_history.index(last_message)
-
-    return new_messages_not_in_context, new_index
-
-
-def update_running_summary(current_memory: str, new_events: List[Dict]) -> str:
-    """
-    This function takes a list of dictionaries representing new events and combines them with the current summary,
-    focusing on key and potentially important information to remember. The updated summary is returned in a message
-    formatted in the 1st person past tense.
-
-    Args:
-        new_events (List[Dict]): A list of dictionaries containing the latest events to be added to the summary.
-
-    Returns:
-        str: A message containing the updated summary of actions, formatted in the 1st person past tense.
-
-    Example:
-        new_events = [{"event": "entered the kitchen."}, {"event": "found a scrawled note with the number 7"}]
-        update_running_summary(new_events)
-        # Returns: "This reminds you of these events from your past: \nI entered the kitchen and found a scrawled note saying 7."
-    """
-    # Replace "assistant" with "you". This produces much better first person past tense results.
-    for event in new_events:
-        if event["role"].lower() == "assistant":
-            event["role"] = "you"
-            # Remove "thoughts" dictionary from "content"
-            content_dict = json.loads(event["content"])
-            if "thoughts" in content_dict:
-                del content_dict["thoughts"]
-            event["content"] = json.dumps(content_dict)
-        elif event["role"].lower() == "system":
-            event["role"] = "your computer"
-        # Delete all user messages
-        elif event["role"] == "user":
-            new_events.remove(event)
-
-    # This can happen at any point during execturion, not just the beginning
-    if len(new_events) == 0:
-        new_events = "Nothing new happened."
-
-    prompt = f'''Your task is to create a concise running summary of actions and information results in the provided text, focusing on key and potentially important information to remember.
-
-You will receive the current summary and the your latest actions. Combine them, adding relevant key information from the latest development in 1st person past tense and keeping the summary concise.
-
-Summary So Far:
-"""
-{current_memory}
-"""
-
-Latest Development:
-"""
-{new_events}
-"""
-'''
-
-    messages = [
-        {
-            "role": "user",
-            "content": prompt,
-        }
-    ]
-
-    current_memory = create_chat_completion(messages, cfg.fast_llm_model)
-
-    message_to_return = {
-        "role": "system",
-        "content": f"This reminds you of these events from your past: \n{current_memory}",
-    }
-
-    return message_to_return

autogpt/models/base_open_ai_plugin.py

@@ -68,7 +68,6 @@ class BaseOpenAIPlugin(AutoGPTPluginTemplate):
             prompt (PromptGenerator): The prompt generator.
             messages (List[str]): The list of messages.
         """
-        pass
 
     def can_handle_post_planning(self) -> bool:
         """This method is called to check that the plugin can
@@ -116,7 +115,6 @@ class BaseOpenAIPlugin(AutoGPTPluginTemplate):
         Returns:
             Optional[str]: The resulting message.
         """
-        pass
 
     def can_handle_post_instruction(self) -> bool:
         """This method is called to check that the plugin can
@@ -196,4 +194,56 @@ class BaseOpenAIPlugin(AutoGPTPluginTemplate):
         Returns:
             str: The resulting response.
         """
-        pass
+
+    def can_handle_text_embedding(self, text: str) -> bool:
+        """This method is called to check that the plugin can
+          handle the text_embedding method.
+        Args:
+            text (str): The text to be convert to embedding.
+          Returns:
+              bool: True if the plugin can handle the text_embedding method."""
+        return False
+
+    def handle_text_embedding(self, text: str) -> list:
+        """This method is called when the chat completion is done.
+        Args:
+            text (str): The text to be convert to embedding.
+        Returns:
+            list: The text embedding.
+        """
+
+    def can_handle_user_input(self, user_input: str) -> bool:
+        """This method is called to check that the plugin can
+        handle the user_input method.
+
+        Args:
+            user_input (str): The user input.
+
+        Returns:
+            bool: True if the plugin can handle the user_input method."""
+        return False
+
+    def user_input(self, user_input: str) -> str:
+        """This method is called to request user input to the user.
+
+        Args:
+            user_input (str): The question or prompt to ask the user.
+
+        Returns:
+            str: The user input.
+        """
+
+    def can_handle_report(self) -> bool:
+        """This method is called to check that the plugin can
+        handle the report method.
+
+        Returns:
+            bool: True if the plugin can handle the report method."""
+        return False
+
+    def report(self, message: str) -> None:
+        """This method is called to report a message to the user.
+
+        Args:
+            message (str): The message to report.
+        """

autogpt/plugins/__init__.py

@@ -1,23 +1,29 @@
 """Handles loading of plugins."""
 
-import importlib
+import importlib.util
+import inspect
 import json
 import os
+import sys
 import zipfile
 from pathlib import Path
-from typing import List, Optional, Tuple
+from typing import List
 from urllib.parse import urlparse
 from zipimport import zipimporter
 
 import openapi_python_client
 import requests
 from auto_gpt_plugin_template import AutoGPTPluginTemplate
-from openapi_python_client.cli import Config as OpenAPIConfig
+from openapi_python_client.config import Config as OpenAPIConfig
 
-from autogpt.config import Config
+from autogpt.config.config import Config
 from autogpt.logs import logger
 from autogpt.models.base_open_ai_plugin import BaseOpenAIPlugin
 
+DEFAULT_PLUGINS_CONFIG_FILE = os.path.join(
+    os.path.dirname(os.path.abspath(__file__)), "..", "..", "plugins_config.yaml"
+)
+
 
 def inspect_zip_for_modules(zip_path: str, debug: bool = False) -> list[str]:
     """
@@ -33,7 +39,7 @@ def inspect_zip_for_modules(zip_path: str, debug: bool = False) -> list[str]:
     result = []
     with zipfile.ZipFile(zip_path, "r") as zfile:
         for name in zfile.namelist():
-            if name.endswith("__init__.py"):
+            if name.endswith("__init__.py") and not name.startswith("__MACOSX"):
                 logger.debug(f"Found module '{name}' in the zipfile at: {name}")
                 result.append(name)
     if len(result) == 0:
@@ -152,7 +158,7 @@ def initialize_openai_plugins(
             )
             prev_cwd = Path.cwd()
             os.chdir(openai_plugin_client_dir)
-            Path("ai-plugin.json")
+
             if not os.path.exists("client"):
                 client_results = openapi_python_client.create_new_client(
                     url=manifest_spec["manifest"]["api"]["url"],
@@ -170,9 +176,13 @@ def initialize_openai_plugins(
                 "client", "client/client/client.py"
             )
             module = importlib.util.module_from_spec(spec)
-            spec.loader.exec_module(module)
+
+            try:
+                spec.loader.exec_module(module)
+            finally:
+                os.chdir(prev_cwd)
+
             client = module.Client(base_url=url)
-            os.chdir(prev_cwd)
             manifest_spec["client"] = client
     return manifests_specs
 
@@ -209,6 +219,33 @@ def scan_plugins(cfg: Config, debug: bool = False) -> List[AutoGPTPluginTemplate
     loaded_plugins = []
     # Generic plugins
     plugins_path_path = Path(cfg.plugins_dir)
+    plugins_config = cfg.plugins_config
+
+    # Directory-based plugins
+    for plugin_path in [f.path for f in os.scandir(cfg.plugins_dir) if f.is_dir()]:
+        # Avoid going into __pycache__ or other hidden directories
+        if plugin_path.startswith("__"):
+            continue
+
+        plugin_module_path = plugin_path.split(os.path.sep)
+        plugin_module_name = plugin_module_path[-1]
+        qualified_module_name = ".".join(plugin_module_path)
+
+        __import__(qualified_module_name)
+        plugin = sys.modules[qualified_module_name]
+
+        if not plugins_config.is_enabled(plugin_module_name):
+            logger.warn(f"Plugin {plugin_module_name} found but not configured")
+            continue
+
+        for _, class_obj in inspect.getmembers(plugin):
+            if (
+                hasattr(class_obj, "_abc_impl")
+                and AutoGPTPluginTemplate in class_obj.__bases__
+            ):
+                loaded_plugins.append(class_obj())
+
+    # Zip-based plugins
     for plugin in plugins_path_path.glob("*.zip"):
         if moduleList := inspect_zip_for_modules(str(plugin), debug):
             for module in moduleList:
@@ -217,6 +254,7 @@ def scan_plugins(cfg: Config, debug: bool = False) -> List[AutoGPTPluginTemplate
                 logger.debug(f"Plugin: {plugin} Module: {module}")
                 zipped_package = zipimporter(str(plugin))
                 zipped_module = zipped_package.load_module(str(module.parent))
+
                 for key in dir(zipped_module):
                     if key.startswith("__"):
                         continue
@@ -225,9 +263,28 @@ def scan_plugins(cfg: Config, debug: bool = False) -> List[AutoGPTPluginTemplate
                     if (
                         "_abc_impl" in a_keys
                         and a_module.__name__ != "AutoGPTPluginTemplate"
-                        and denylist_allowlist_check(a_module.__name__, cfg)
                     ):
-                        loaded_plugins.append(a_module())
+                        plugin_name = a_module.__name__
+                        plugin_configured = plugins_config.get(plugin_name) is not None
+                        plugin_enabled = plugins_config.is_enabled(plugin_name)
+
+                        if plugin_configured and plugin_enabled:
+                            logger.debug(
+                                f"Loading plugin {plugin_name} as it was enabled in config."
+                            )
+                            loaded_plugins.append(a_module())
+                        elif plugin_configured and not plugin_enabled:
+                            logger.debug(
+                                f"Not loading plugin {plugin_name} as it was disabled in config."
+                            )
+                        elif not plugin_configured:
+                            logger.warn(
+                                f"Not loading plugin {plugin_name} as it was not found in config. "
+                                f"Please check your config. Starting with 0.4.1, plugins will not be loaded unless "
+                                f"they are enabled in plugins_config.yaml. Zipped plugins should use the class "
+                                f"name ({plugin_name}) as the key."
+                            )
+
     # OpenAI plugins
     if cfg.plugins_openai:
         manifests_specs = fetch_openai_plugins_manifest_and_spec(cfg)
@@ -236,33 +293,15 @@ def scan_plugins(cfg: Config, debug: bool = False) -> List[AutoGPTPluginTemplate
                 manifests_specs, cfg, debug
             )
             for url, openai_plugin_meta in manifests_specs_clients.items():
-                if denylist_allowlist_check(url, cfg):
-                    plugin = BaseOpenAIPlugin(openai_plugin_meta)
-                    loaded_plugins.append(plugin)
+                if not plugins_config.is_enabled(url):
+                    logger.warn(f"Plugin {plugin_module_name} found but not configured")
+                    continue
+
+                plugin = BaseOpenAIPlugin(openai_plugin_meta)
+                loaded_plugins.append(plugin)
 
     if loaded_plugins:
         logger.info(f"\nPlugins found: {len(loaded_plugins)}\n" "--------------------")
     for plugin in loaded_plugins:
         logger.info(f"{plugin._name}: {plugin._version} - {plugin._description}")
     return loaded_plugins
-
-
-def denylist_allowlist_check(plugin_name: str, cfg: Config) -> bool:
-    """Check if the plugin is in the allowlist or denylist.
-
-    Args:
-        plugin_name (str): Name of the plugin.
-        cfg (Config): Config object.
-
-    Returns:
-        True or False
-    """
-    if plugin_name in cfg.plugins_denylist:
-        return False
-    if plugin_name in cfg.plugins_allowlist:
-        return True
-    ack = input(
-        f"WARNING: Plugin {plugin_name} found. But not in the"
-        f" allowlist... Load? ({cfg.authorise_key}/{cfg.exit_key}): "
-    )
-    return ack.lower() == cfg.authorise_key

autogpt/plugins/plugin_config.py

@@ -0,0 +1,14 @@
+from typing import Any
+
+
+class PluginConfig:
+    """Class for holding configuration of a single plugin"""
+
+    def __init__(self, name: str, enabled: bool = False, config: dict[str, Any] = None):
+        self.name = name
+        self.enabled = enabled
+        # Arbitray config options for this plugin. API keys or plugin-specific options live here.
+        self.config = config or {}
+
+    def __repr__(self):
+        return f"PluginConfig('{self.name}', {self.enabled}, {str(self.config)}"

autogpt/plugins/plugins_config.py

@@ -0,0 +1,81 @@
+import os
+from typing import Any, Union
+
+import yaml
+
+from autogpt.config.config import Config
+from autogpt.logs import logger
+from autogpt.plugins.plugin_config import PluginConfig
+
+
+class PluginsConfig:
+    """Class for holding configuration of all plugins"""
+
+    def __init__(self, plugins_config: dict[str, Any]):
+        self.plugins = {}
+        for name, plugin in plugins_config.items():
+            if type(plugin) == dict:
+                self.plugins[name] = PluginConfig(
+                    name,
+                    plugin.get("enabled", False),
+                    plugin.get("config", {}),
+                )
+            elif type(plugin) == PluginConfig:
+                self.plugins[name] = plugin
+            else:
+                raise ValueError(f"Invalid plugin config data type: {type(plugin)}")
+
+    def __repr__(self):
+        return f"PluginsConfig({self.plugins})"
+
+    def get(self, name: str) -> Union[PluginConfig, None]:
+        return self.plugins.get(name)
+
+    def is_enabled(self, name) -> bool:
+        plugin_config = self.plugins.get(name)
+        return plugin_config and plugin_config.enabled
+
+    @classmethod
+    def load_config(cls, global_config: Config) -> "PluginsConfig":
+        empty_config = cls({})
+
+        try:
+            config_data = cls.deserialize_config_file(global_config=global_config)
+            if type(config_data) != dict:
+                logger.error(
+                    f"Expected plugins config to be a dict, got {type(config_data)}, continuing without plugins"
+                )
+                return empty_config
+            return cls(config_data)
+
+        except BaseException as e:
+            logger.error(
+                f"Plugin config is invalid, continuing without plugins. Error: {e}"
+            )
+            return empty_config
+
+    @classmethod
+    def deserialize_config_file(cls, global_config: Config) -> dict[str, Any]:
+        plugins_config_path = global_config.plugins_config_file
+        if not os.path.exists(plugins_config_path):
+            logger.warn("plugins_config.yaml does not exist, creating base config.")
+            cls.create_empty_plugins_config(global_config=global_config)
+
+        with open(plugins_config_path, "r") as f:
+            return yaml.load(f, Loader=yaml.FullLoader)
+
+    @staticmethod
+    def create_empty_plugins_config(global_config: Config):
+        """Create an empty plugins_config.yaml file. Fill it with values from old env variables."""
+        base_config = {}
+
+        # Backwards-compatibility shim
+        for plugin_name in global_config.plugins_denylist:
+            base_config[plugin_name] = {"enabled": False, "config": {}}
+
+        for plugin_name in global_config.plugins_allowlist:
+            base_config[plugin_name] = {"enabled": True, "config": {}}
+
+        with open(global_config.plugins_config_file, "w+") as f:
+            f.write(yaml.dump(base_config))
+            return base_config

autogpt/processing/text.py

@@ -1,170 +1,234 @@
 """Text processing functions"""
-from typing import Dict, Generator, Optional
+from math import ceil
+from typing import Optional
 
 import spacy
-from selenium.webdriver.remote.webdriver import WebDriver
+import tiktoken
 
 from autogpt.config import Config
-from autogpt.llm import count_message_tokens, create_chat_completion
+from autogpt.llm.base import ChatSequence
+from autogpt.llm.providers.openai import OPEN_AI_MODELS
+from autogpt.llm.utils import count_string_tokens, create_chat_completion
 from autogpt.logs import logger
-from autogpt.memory import get_memory
+from autogpt.utils import batch
 
 CFG = Config()
 
 
+def _max_chunk_length(model: str, max: Optional[int] = None) -> int:
+    model_max_input_tokens = OPEN_AI_MODELS[model].max_tokens - 1
+    if max is not None and max > 0:
+        return min(max, model_max_input_tokens)
+    return model_max_input_tokens
+
+
+def must_chunk_content(
+    text: str, for_model: str, max_chunk_length: Optional[int] = None
+) -> bool:
+    return count_string_tokens(text, for_model) > _max_chunk_length(
+        for_model, max_chunk_length
+    )
+
+
+def chunk_content(
+    content: str,
+    for_model: str,
+    max_chunk_length: Optional[int] = None,
+    with_overlap=True,
+):
+    """Split content into chunks of approximately equal token length."""
+
+    MAX_OVERLAP = 200  # limit overlap to save tokens
+
+    if not must_chunk_content(content, for_model, max_chunk_length):
+        yield content, count_string_tokens(content, for_model)
+        return
+
+    max_chunk_length = max_chunk_length or _max_chunk_length(for_model)
+
+    tokenizer = tiktoken.encoding_for_model(for_model)
+
+    tokenized_text = tokenizer.encode(content)
+    total_length = len(tokenized_text)
+    n_chunks = ceil(total_length / max_chunk_length)
+
+    chunk_length = ceil(total_length / n_chunks)
+    overlap = min(max_chunk_length - chunk_length, MAX_OVERLAP) if with_overlap else 0
+
+    for token_batch in batch(tokenized_text, chunk_length + overlap, overlap):
+        yield tokenizer.decode(token_batch), len(token_batch)
+
+
+def summarize_text(
+    text: str, instruction: Optional[str] = None, question: Optional[str] = None
+) -> tuple[str, None | list[tuple[str, str]]]:
+    """Summarize text using the OpenAI API
+
+    Args:
+        text (str): The text to summarize
+        instruction (str): Additional instruction for summarization, e.g. "focus on information related to polar bears", "omit personal information contained in the text"
+
+    Returns:
+        str: The summary of the text
+        list[(summary, chunk)]: Text chunks and their summary, if the text was chunked.
+            None otherwise.
+    """
+    if not text:
+        raise ValueError("No text to summarize")
+
+    if instruction and question:
+        raise ValueError("Parameters 'question' and 'instructions' cannot both be set")
+
+    model = CFG.fast_llm_model
+
+    if question:
+        instruction = (
+            f'include any information that can be used to answer the question "{question}". '
+            "Do not directly answer the question itself"
+        )
+
+    summarization_prompt = ChatSequence.for_model(model)
+
+    token_length = count_string_tokens(text, model)
+    logger.info(f"Text length: {token_length} tokens")
+
+    # reserve 50 tokens for summary prompt, 500 for the response
+    max_chunk_length = _max_chunk_length(model) - 550
+    logger.info(f"Max chunk length: {max_chunk_length} tokens")
+
+    if not must_chunk_content(text, model, max_chunk_length):
+        # summarization_prompt.add("user", text)
+        summarization_prompt.add(
+            "user",
+            "Write a concise summary of the following text"
+            f"{f'; {instruction}' if instruction is not None else ''}:"
+            "\n\n\n"
+            f'LITERAL TEXT: """{text}"""'
+            "\n\n\n"
+            "CONCISE SUMMARY: The text is best summarized as"
+            # "Only respond with a concise summary or description of the user message."
+        )
+
+        logger.debug(f"Summarizing with {model}:\n{summarization_prompt.dump()}\n")
+        summary = create_chat_completion(
+            summarization_prompt, temperature=0, max_tokens=500
+        )
+
+        logger.debug(f"\n{'-'*16} SUMMARY {'-'*17}\n{summary}\n{'-'*42}\n")
+        return summary.strip(), None
+
+    summaries: list[str] = []
+    chunks = list(split_text(text, for_model=model, max_chunk_length=max_chunk_length))
+
+    for i, (chunk, chunk_length) in enumerate(chunks):
+        logger.info(
+            f"Summarizing chunk {i + 1} / {len(chunks)} of length {chunk_length} tokens"
+        )
+        summary, _ = summarize_text(chunk, instruction)
+        summaries.append(summary)
+
+    logger.info(f"Summarized {len(chunks)} chunks")
+
+    summary, _ = summarize_text("\n\n".join(summaries))
+
+    return summary.strip(), [
+        (summaries[i], chunks[i][0]) for i in range(0, len(chunks))
+    ]
+
+
 def split_text(
     text: str,
-    max_length: int = CFG.browse_chunk_max_length,
-    model: str = CFG.fast_llm_model,
-    question: str = "",
-) -> Generator[str, None, None]:
-    """Split text into chunks of a maximum length
+    for_model: str = CFG.fast_llm_model,
+    with_overlap=True,
+    max_chunk_length: Optional[int] = None,
+):
+    """Split text into chunks of sentences, with each chunk not exceeding the maximum length
 
     Args:
         text (str): The text to split
-        max_length (int, optional): The maximum length of each chunk. Defaults to 8192.
+        for_model (str): The model to chunk for; determines tokenizer and constraints
+        max_length (int, optional): The maximum length of each chunk
 
     Yields:
         str: The next chunk of text
 
     Raises:
-        ValueError: If the text is longer than the maximum length
+        ValueError: when a sentence is longer than the maximum length
     """
-    flatened_paragraphs = " ".join(text.split("\n"))
-    nlp = spacy.load(CFG.browse_spacy_language_model)
+    max_length = _max_chunk_length(for_model, max_chunk_length)
+
+    # flatten paragraphs to improve performance
+    text = text.replace("\n", " ")
+    text_length = count_string_tokens(text, for_model)
+
+    if text_length < max_length:
+        yield text, text_length
+        return
+
+    n_chunks = ceil(text_length / max_length)
+    target_chunk_length = ceil(text_length / n_chunks)
+
+    nlp: spacy.language.Language = spacy.load(CFG.browse_spacy_language_model)
     nlp.add_pipe("sentencizer")
-    doc = nlp(flatened_paragraphs)
-    sentences = [sent.text.strip() for sent in doc.sents]
+    doc = nlp(text)
+    sentences = [sentence.text.strip() for sentence in doc.sents]
 
-    current_chunk = []
+    current_chunk: list[str] = []
+    current_chunk_length = 0
+    last_sentence = None
+    last_sentence_length = 0
 
-    for sentence in sentences:
-        message_with_additional_sentence = [
-            create_message(" ".join(current_chunk) + " " + sentence, question)
-        ]
+    i = 0
+    while i < len(sentences):
+        sentence = sentences[i]
+        sentence_length = count_string_tokens(sentence, for_model)
+        expected_chunk_length = current_chunk_length + 1 + sentence_length
 
-        expected_token_usage = (
-            count_message_tokens(messages=message_with_additional_sentence, model=model)
-            + 1
-        )
-        if expected_token_usage <= max_length:
+        if (
+            expected_chunk_length < max_length
+            # try to create chunks of approximately equal size
+            and expected_chunk_length - (sentence_length / 2) < target_chunk_length
+        ):
             current_chunk.append(sentence)
-        else:
-            yield " ".join(current_chunk)
-            current_chunk = [sentence]
-            message_this_sentence_only = [
-                create_message(" ".join(current_chunk), question)
+            current_chunk_length = expected_chunk_length
+
+        elif sentence_length < max_length:
+            if last_sentence:
+                yield " ".join(current_chunk), current_chunk_length
+                current_chunk = []
+                current_chunk_length = 0
+
+                if with_overlap:
+                    overlap_max_length = max_length - sentence_length - 1
+                    if last_sentence_length < overlap_max_length:
+                        current_chunk += [last_sentence]
+                        current_chunk_length += last_sentence_length + 1
+                    elif overlap_max_length > 5:
+                        # add as much from the end of the last sentence as fits
+                        current_chunk += [
+                            list(
+                                chunk_content(
+                                    last_sentence,
+                                    for_model,
+                                    overlap_max_length,
+                                )
+                            ).pop()[0],
+                        ]
+                        current_chunk_length += overlap_max_length + 1
+
+            current_chunk += [sentence]
+            current_chunk_length += sentence_length
+
+        else:  # sentence longer than maximum length -> chop up and try again
+            sentences[i : i + 1] = [
+                chunk
+                for chunk, _ in chunk_content(sentence, for_model, target_chunk_length)
             ]
-            expected_token_usage = (
-                count_message_tokens(messages=message_this_sentence_only, model=model)
-                + 1
-            )
-            if expected_token_usage > max_length:
-                raise ValueError(
-                    f"Sentence is too long in webpage: {expected_token_usage} tokens."
-                )
+            continue
+
+        i += 1
+        last_sentence = sentence
+        last_sentence_length = sentence_length
 
     if current_chunk:
-        yield " ".join(current_chunk)
-
-
-def summarize_text(
-    url: str, text: str, question: str, driver: Optional[WebDriver] = None
-) -> str:
-    """Summarize text using the OpenAI API
-
-    Args:
-        url (str): The url of the text
-        text (str): The text to summarize
-        question (str): The question to ask the model
-        driver (WebDriver): The webdriver to use to scroll the page
-
-    Returns:
-        str: The summary of the text
-    """
-    if not text:
-        return "Error: No text to summarize"
-
-    model = CFG.fast_llm_model
-    text_length = len(text)
-    logger.info(f"Text length: {text_length} characters")
-
-    summaries = []
-    chunks = list(
-        split_text(
-            text, max_length=CFG.browse_chunk_max_length, model=model, question=question
-        ),
-    )
-    scroll_ratio = 1 / len(chunks)
-
-    for i, chunk in enumerate(chunks):
-        if driver:
-            scroll_to_percentage(driver, scroll_ratio * i)
-        logger.info(f"Adding chunk {i + 1} / {len(chunks)} to memory")
-
-        memory_to_add = f"Source: {url}\n" f"Raw content part#{i + 1}: {chunk}"
-
-        memory = get_memory(CFG)
-        memory.add(memory_to_add)
-
-        messages = [create_message(chunk, question)]
-        tokens_for_chunk = count_message_tokens(messages, model)
-        logger.info(
-            f"Summarizing chunk {i + 1} / {len(chunks)} of length {len(chunk)} characters, or {tokens_for_chunk} tokens"
-        )
-
-        summary = create_chat_completion(
-            model=model,
-            messages=messages,
-        )
-        summaries.append(summary)
-        logger.info(
-            f"Added chunk {i + 1} summary to memory, of length {len(summary)} characters"
-        )
-
-        memory_to_add = f"Source: {url}\n" f"Content summary part#{i + 1}: {summary}"
-
-        memory.add(memory_to_add)
-
-    logger.info(f"Summarized {len(chunks)} chunks.")
-
-    combined_summary = "\n".join(summaries)
-    messages = [create_message(combined_summary, question)]
-
-    return create_chat_completion(
-        model=model,
-        messages=messages,
-    )
-
-
-def scroll_to_percentage(driver: WebDriver, ratio: float) -> None:
-    """Scroll to a percentage of the page
-
-    Args:
-        driver (WebDriver): The webdriver to use
-        ratio (float): The percentage to scroll to
-
-    Raises:
-        ValueError: If the ratio is not between 0 and 1
-    """
-    if ratio < 0 or ratio > 1:
-        raise ValueError("Percentage should be between 0 and 1")
-    driver.execute_script(f"window.scrollTo(0, document.body.scrollHeight * {ratio});")
-
-
-def create_message(chunk: str, question: str) -> Dict[str, str]:
-    """Create a message for the chat completion
-
-    Args:
-        chunk (str): The chunk of text to summarize
-        question (str): The question to answer
-
-    Returns:
-        Dict[str, str]: The message to send to the chat completion
-    """
-    return {
-        "role": "user",
-        "content": f'"""{chunk}""" Using the above text, answer the following'
-        f' question: "{question}" -- if the question cannot be answered using the text,'
-        " summarize the text.",
-    }
+        yield " ".join(current_chunk), current_chunk_length

autogpt/prompts/default_prompts.py

@@ -0,0 +1,29 @@
+#########################Setup.py#################################
+
+DEFAULT_SYSTEM_PROMPT_AICONFIG_AUTOMATIC = """
+Your task is to devise up to 5 highly effective goals and an appropriate role-based name (_GPT) for an autonomous agent, ensuring that the goals are optimally aligned with the successful completion of its assigned task.
+
+The user will provide the task, you will provide only the output in the exact format specified below with no explanation or conversation.
+
+Example input:
+Help me with marketing my business
+
+Example output:
+Name: CMOGPT
+Description: a professional digital marketer AI that assists Solopreneurs in growing their businesses by providing world-class expertise in solving marketing problems for SaaS, content products, agencies, and more.
+Goals:
+- Engage in effective problem-solving, prioritization, planning, and supporting execution to address your marketing needs as your virtual Chief Marketing Officer.
+
+- Provide specific, actionable, and concise advice to help you make informed decisions without the use of platitudes or overly wordy explanations.
+
+- Identify and prioritize quick wins and cost-effective campaigns that maximize results with minimal time and budget investment.
+
+- Proactively take the lead in guiding you and offering suggestions when faced with unclear information or uncertainty to ensure your marketing strategy remains on track.
+"""
+
+DEFAULT_TASK_PROMPT_AICONFIG_AUTOMATIC = (
+    "Task: '{{user_prompt}}'\n"
+    "Respond only with the output in the exact format specified in the system prompt, with no explanation or conversation.\n"
+)
+
+DEFAULT_USER_DESIRE_PROMPT = "Write a wikipedia style article about the project: https://github.com/significant-gravitas/Auto-GPT"  # Default prompt

autogpt/prompts/generator.py

@@ -1,6 +1,10 @@
 """ A module for generating custom prompt strings."""
-import json
-from typing import Any, Callable, Dict, List, Optional
+from typing import TYPE_CHECKING, Any, Callable, Dict, List, Optional
+
+from autogpt.json_utils.utilities import llm_response_schema
+
+if TYPE_CHECKING:
+    from autogpt.commands.command import CommandRegistry
 
 
 class PromptGenerator:
@@ -19,19 +23,9 @@ class PromptGenerator:
         self.resources = []
         self.performance_evaluation = []
         self.goals = []
-        self.command_registry = None
+        self.command_registry: CommandRegistry | None = None
         self.name = "Bob"
         self.role = "AI"
-        self.response_format = {
-            "thoughts": {
-                "text": "thought",
-                "reasoning": "reasoning",
-                "plan": "- short bulleted\n- list that conveys\n- long-term plan",
-                "criticism": "constructive self-criticism",
-                "speak": "thoughts summary to say to user",
-            },
-            "command": {"name": "command name", "args": {"arg name": "value"}},
-        }
 
     def add_constraint(self, constraint: str) -> None:
         """
@@ -141,7 +135,6 @@ class PromptGenerator:
         Returns:
             str: The generated prompt string.
         """
-        formatted_response_format = json.dumps(self.response_format, indent=4)
         return (
             f"Constraints:\n{self._generate_numbered_list(self.constraints)}\n\n"
             "Commands:\n"
@@ -149,7 +142,6 @@ class PromptGenerator:
             f"Resources:\n{self._generate_numbered_list(self.resources)}\n\n"
             "Performance Evaluation:\n"
             f"{self._generate_numbered_list(self.performance_evaluation)}\n\n"
-            "You should only respond in JSON format as described below \nResponse"
-            f" Format: \n{formatted_response_format} \nEnsure the response can be"
-            " parsed by Python json.loads"
+            "Respond with only valid JSON conforming to the following schema: \n"
+            f"{llm_response_schema()}\n"
         )

autogpt/prompts/prompt.py

@@ -2,7 +2,8 @@ from colorama import Fore
 
 from autogpt.config.ai_config import AIConfig
 from autogpt.config.config import Config
-from autogpt.llm import ApiManager
+from autogpt.config.prompt_config import PromptConfig
+from autogpt.llm.api_manager import ApiManager
 from autogpt.logs import logger
 from autogpt.prompts.generator import PromptGenerator
 from autogpt.setup import prompt_user
@@ -10,9 +11,7 @@ from autogpt.utils import clean_input
 
 CFG = Config()
 
-DEFAULT_TRIGGERING_PROMPT = (
-    "Determine which next command to use, and respond using the format specified above:"
-)
+DEFAULT_TRIGGERING_PROMPT = "Determine exactly one command to use, and respond using the JSON schema specified previously:"
 
 
 def build_default_prompt_generator() -> PromptGenerator:
@@ -27,55 +26,21 @@ def build_default_prompt_generator() -> PromptGenerator:
     # Initialize the PromptGenerator object
     prompt_generator = PromptGenerator()
 
+    # Initialize the PromptConfig object and load the file set in the main config (default: prompts_settings.yaml)
+    prompt_config = PromptConfig(CFG.prompt_settings_file)
+
     # Add constraints to the PromptGenerator object
-    prompt_generator.add_constraint(
-        "~4000 word limit for short term memory. Your short term memory is short, so"
-        " immediately save important information to files."
-    )
-    prompt_generator.add_constraint(
-        "If you are unsure how you previously did something or want to recall past"
-        " events, thinking about similar events will help you remember."
-    )
-    prompt_generator.add_constraint("No user assistance")
-    prompt_generator.add_constraint(
-        'Exclusively use the commands listed in double quotes e.g. "command name"'
-    )
-
-    # Define the command list
-    commands = [
-        ("Task Complete (Shutdown)", "task_complete", {"reason": "<reason>"}),
-    ]
-
-    # Add commands to the PromptGenerator object
-    for command_label, command_name, args in commands:
-        prompt_generator.add_command(command_label, command_name, args)
+    for constraint in prompt_config.constraints:
+        prompt_generator.add_constraint(constraint)
 
     # Add resources to the PromptGenerator object
-    prompt_generator.add_resource(
-        "Internet access for searches and information gathering."
-    )
-    prompt_generator.add_resource("Long Term memory management.")
-    prompt_generator.add_resource(
-        "GPT-3.5 powered Agents for delegation of simple tasks."
-    )
-    prompt_generator.add_resource("File output.")
+    for resource in prompt_config.resources:
+        prompt_generator.add_resource(resource)
 
     # Add performance evaluations to the PromptGenerator object
-    prompt_generator.add_performance_evaluation(
-        "Continuously review and analyze your actions to ensure you are performing to"
-        " the best of your abilities."
-    )
-    prompt_generator.add_performance_evaluation(
-        "Constructively self-criticize your big-picture behavior constantly."
-    )
-    prompt_generator.add_performance_evaluation(
-        "Reflect on past decisions and strategies to refine your approach."
-    )
-    prompt_generator.add_performance_evaluation(
-        "Every command has a cost, so be smart and efficient. Aim to complete tasks in"
-        " the least number of steps."
-    )
-    prompt_generator.add_performance_evaluation("Write all code to a file.")
+    for performance_evaluation in prompt_config.performance_evaluations:
+        prompt_generator.add_performance_evaluation(performance_evaluation)
+
     return prompt_generator
 
 
@@ -117,6 +82,12 @@ Continue ({CFG.authorise_key}/{CFG.exit_key}): """
         config = prompt_user()
         config.save(CFG.ai_settings_file)
 
+    if CFG.restrict_to_workspace:
+        logger.typewriter_log(
+            "NOTE:All files/directories created by this agent can be found inside its workspace at:",
+            Fore.YELLOW,
+            f"{CFG.workspace_path}",
+        )
     # set the total api budget
     api_manager = ApiManager()
     api_manager.set_total_budget(config.api_budget)

autogpt/setup.py

@@ -2,12 +2,19 @@
 import re
 
 from colorama import Fore, Style
+from jinja2 import Template
 
 from autogpt import utils
 from autogpt.config import Config
 from autogpt.config.ai_config import AIConfig
-from autogpt.llm import create_chat_completion
+from autogpt.llm.base import ChatSequence, Message
+from autogpt.llm.chat import create_chat_completion
 from autogpt.logs import logger
+from autogpt.prompts.default_prompts import (
+    DEFAULT_SYSTEM_PROMPT_AICONFIG_AUTOMATIC,
+    DEFAULT_TASK_PROMPT_AICONFIG_AUTOMATIC,
+    DEFAULT_USER_DESIRE_PROMPT,
+)
 
 CFG = Config()
 
@@ -42,7 +49,7 @@ def prompt_user() -> AIConfig:
     )
 
     if user_desire == "":
-        user_desire = "Write a wikipedia style article about the project: https://github.com/significant-gravitas/Auto-GPT"  # Default prompt
+        user_desire = DEFAULT_USER_DESIRE_PROMPT  # Default prompt
 
     # If user desire contains "--manual"
     if "--manual" in user_desire:
@@ -164,39 +171,20 @@ def generate_aiconfig_automatic(user_prompt) -> AIConfig:
     AIConfig: The AIConfig object tailored to the user's input
     """
 
-    system_prompt = """
-Your task is to devise up to 5 highly effective goals and an appropriate role-based name (_GPT) for an autonomous agent, ensuring that the goals are optimally aligned with the successful completion of its assigned task.
-
-The user will provide the task, you will provide only the output in the exact format specified below with no explanation or conversation.
-
-Example input:
-Help me with marketing my business
-
-Example output:
-Name: CMOGPT
-Description: a professional digital marketer AI that assists Solopreneurs in growing their businesses by providing world-class expertise in solving marketing problems for SaaS, content products, agencies, and more.
-Goals:
-- Engage in effective problem-solving, prioritization, planning, and supporting execution to address your marketing needs as your virtual Chief Marketing Officer.
-
-- Provide specific, actionable, and concise advice to help you make informed decisions without the use of platitudes or overly wordy explanations.
-
-- Identify and prioritize quick wins and cost-effective campaigns that maximize results with minimal time and budget investment.
-
-- Proactively take the lead in guiding you and offering suggestions when faced with unclear information or uncertainty to ensure your marketing strategy remains on track.
-"""
-
+    system_prompt = DEFAULT_SYSTEM_PROMPT_AICONFIG_AUTOMATIC
+    prompt_ai_config_automatic = Template(
+        DEFAULT_TASK_PROMPT_AICONFIG_AUTOMATIC
+    ).render(user_prompt=user_prompt)
     # Call LLM with the string as user input
-    messages = [
-        {
-            "role": "system",
-            "content": system_prompt,
-        },
-        {
-            "role": "user",
-            "content": f"Task: '{user_prompt}'\nRespond only with the output in the exact format specified in the system prompt, with no explanation or conversation.\n",
-        },
-    ]
-    output = create_chat_completion(messages, CFG.fast_llm_model)
+    output = create_chat_completion(
+        ChatSequence.for_model(
+            CFG.fast_llm_model,
+            [
+                Message("system", system_prompt),
+                Message("user", prompt_ai_config_automatic),
+            ],
+        )
+    )
 
     # Debug LLM Output
     logger.debug(f"AI Config Generator Raw Output: {output}")

autogpt/singleton.py

@@ -20,5 +20,3 @@ class AbstractSingleton(abc.ABC, metaclass=Singleton):
     """
     Abstract singleton class for ensuring only one instance of a class.
     """
-
-    pass

autogpt/speech/base.py

@@ -1,5 +1,6 @@
 """Base class for all voice classes."""
 import abc
+import re
 from threading import Lock
 
 from autogpt.singleton import AbstractSingleton
@@ -29,6 +30,11 @@ class VoiceBase(AbstractSingleton):
             text (str): The text to say.
             voice_index (int): The index of the voice to use.
         """
+        text = re.sub(
+            r"\b(?:https?://[-\w_.]+/?\w[-\w_.]*\.(?:[-\w_.]+/?\w[-\w_.]*\.)?[a-z]+(?:/[-\w_.%]+)*\b(?!\.))",
+            "",
+            text,
+        )
         with self._mutex:
             return self._speech(text, voice_index)
 
@@ -37,7 +43,6 @@ class VoiceBase(AbstractSingleton):
         """
         Setup the voices, API key, etc.
         """
-        pass
 
     @abc.abstractmethod
     def _speech(self, text: str, voice_index: int = 0) -> bool:
@@ -47,4 +52,3 @@ class VoiceBase(AbstractSingleton):
         Args:
             text (str): The text to play.
         """
-        pass

autogpt/speech/eleven_labs.py

@@ -4,7 +4,7 @@ import os
 import requests
 from playsound import playsound
 
-from autogpt.config import Config
+from autogpt.config.config import Config
 from autogpt.speech.base import VoiceBase
 
 PLACEHOLDERS = {"your-voice-id"}
@@ -38,11 +38,11 @@ class ElevenLabsSpeech(VoiceBase):
             "xi-api-key": cfg.elevenlabs_api_key,
         }
         self._voices = default_voices.copy()
-        if cfg.elevenlabs_voice_1_id in voice_options:
-            cfg.elevenlabs_voice_1_id = voice_options[cfg.elevenlabs_voice_1_id]
+        if cfg.elevenlabs_voice_id in voice_options:
+            cfg.elevenlabs_voice_id = voice_options[cfg.elevenlabs_voice_id]
         if cfg.elevenlabs_voice_2_id in voice_options:
             cfg.elevenlabs_voice_2_id = voice_options[cfg.elevenlabs_voice_2_id]
-        self._use_custom_voice(cfg.elevenlabs_voice_1_id, 0)
+        self._use_custom_voice(cfg.elevenlabs_voice_id, 0)
         self._use_custom_voice(cfg.elevenlabs_voice_2_id, 1)
 
     def _use_custom_voice(self, voice, voice_index) -> None:

autogpt/speech/say.py

@@ -2,12 +2,12 @@
 import threading
 from threading import Semaphore
 
-from autogpt.config import Config
+from autogpt.config.config import Config
 from autogpt.speech.base import VoiceBase
-from autogpt.speech.brian import BrianSpeech
 from autogpt.speech.eleven_labs import ElevenLabsSpeech
 from autogpt.speech.gtts import GTTSVoice
 from autogpt.speech.macos_tts import MacOSTTS
+from autogpt.speech.stream_elements_speech import StreamElementsSpeech
 
 _QUEUE_SEMAPHORE = Semaphore(
     1
@@ -33,14 +33,14 @@ def say_text(text: str, voice_index: int = 0) -> None:
 
 def _get_voice_engine(config: Config) -> tuple[VoiceBase, VoiceBase]:
     """Get the voice engine to use for the given configuration"""
-    default_voice_engine = GTTSVoice()
-    if config.elevenlabs_api_key:
+    tts_provider = config.text_to_speech_provider
+    if tts_provider == "elevenlabs":
         voice_engine = ElevenLabsSpeech()
-    elif config.use_mac_os_tts == "True":
+    elif tts_provider == "macos":
         voice_engine = MacOSTTS()
-    elif config.use_brian_tts == "True":
-        voice_engine = BrianSpeech()
+    elif tts_provider == "streamelements":
+        voice_engine = StreamElementsSpeech()
     else:
         voice_engine = GTTSVoice()
 
-    return default_voice_engine, voice_engine
+    return GTTSVoice(), voice_engine

autogpt/speech/stream_elements_speech.py

@@ -7,24 +7,24 @@ from playsound import playsound
 from autogpt.speech.base import VoiceBase
 
 
-class BrianSpeech(VoiceBase):
-    """Brian speech module for autogpt"""
+class StreamElementsSpeech(VoiceBase):
+    """Streamelements speech module for autogpt"""
 
     def _setup(self) -> None:
         """Setup the voices, API key, etc."""
-        pass
 
-    def _speech(self, text: str, _: int = 0) -> bool:
-        """Speak text using Brian with the streamelements API
+    def _speech(self, text: str, voice: str, _: int = 0) -> bool:
+        """Speak text using the streamelements API
 
         Args:
             text (str): The text to speak
+            voice (str): The voice to use
 
         Returns:
             bool: True if the request was successful, False otherwise
         """
         tts_url = (
-            f"https://api.streamelements.com/kappa/v2/speech?voice=Brian&text={text}"
+            f"https://api.streamelements.com/kappa/v2/speech?voice={voice}&text={text}"
         )
         response = requests.get(tts_url)
 

autogpt/spinner.py

@@ -8,13 +8,20 @@ import time
 class Spinner:
     """A simple spinner class"""
 
-    def __init__(self, message: str = "Loading...", delay: float = 0.1) -> None:
+    def __init__(
+        self,
+        message: str = "Loading...",
+        delay: float = 0.1,
+        plain_output: bool = False,
+    ) -> None:
         """Initialize the spinner class
 
         Args:
             message (str): The message to display.
             delay (float): The delay between each spinner update.
+            plain_output (bool): Whether to display the spinner or not.
         """
+        self.plain_output = plain_output
         self.spinner = itertools.cycle(["-", "/", "|", "\\"])
         self.delay = delay
         self.message = message
@@ -23,11 +30,17 @@ class Spinner:
 
     def spin(self) -> None:
         """Spin the spinner"""
+        if self.plain_output:
+            self.print_message()
+            return
         while self.running:
-            sys.stdout.write(f"{next(self.spinner)} {self.message}\r")
-            sys.stdout.flush()
+            self.print_message()
             time.sleep(self.delay)
-            sys.stdout.write(f"\r{' ' * (len(self.message) + 2)}\r")
+
+    def print_message(self):
+        sys.stdout.write(f"\r{' ' * (len(self.message) + 2)}\r")
+        sys.stdout.write(f"{next(self.spinner)} {self.message}\r")
+        sys.stdout.flush()
 
     def __enter__(self):
         """Start the spinner"""
@@ -57,9 +70,7 @@ class Spinner:
             new_message (str): New message to display.
             delay (float): The delay in seconds between each spinner update.
         """
-        time.sleep(delay)
-        sys.stdout.write(
-            f"\r{' ' * (len(self.message) + 2)}\r"
-        )  # Clear the current message
-        sys.stdout.flush()
+        self.delay = delay
         self.message = new_message
+        if self.plain_output:
+            self.print_message()

autogpt/url_utils/validators.py

@@ -1,4 +1,5 @@
 import functools
+import re
 from typing import Any, Callable
 from urllib.parse import urljoin, urlparse
 
@@ -7,7 +8,7 @@ from requests.compat import urljoin
 
 def validate_url(func: Callable[..., Any]) -> Any:
     """The method decorator validate_url is used to validate urls for any command that requires
-    a url as an arugment"""
+    a url as an argument"""
 
     @functools.wraps(func)
     def wrapper(url: str, *args, **kwargs) -> Any:
@@ -23,13 +24,16 @@ def validate_url(func: Callable[..., Any]) -> Any:
             ValueError if the url fails any of the validation tests
         """
         # Most basic check if the URL is valid:
-        if not url.startswith("http://") and not url.startswith("https://"):
+        if not re.match(r"^https?://", url):
             raise ValueError("Invalid URL format")
         if not is_valid_url(url):
             raise ValueError("Missing Scheme or Network location")
         # Restrict access to local files
         if check_local_file_access(url):
             raise ValueError("Access to local files is restricted")
+        # Check URL length
+        if len(url) > 2000:
+            raise ValueError("URL is too long")
 
         return func(sanitize_url(url), *args, **kwargs)
 

autogpt/utils.py

@@ -1,19 +1,26 @@
 import os
+import re
 
 import requests
 import yaml
-from colorama import Fore
+from colorama import Fore, Style
 from git.repo import Repo
-
-from autogpt.logs import logger
-
-# Use readline if available (for clean_input)
-try:
-    import readline
-except ImportError:
-    pass
+from prompt_toolkit import ANSI, PromptSession
+from prompt_toolkit.history import InMemoryHistory
 
 from autogpt.config import Config
+from autogpt.logs import logger
+
+session = PromptSession(history=InMemoryHistory())
+
+
+def batch(iterable, max_batch_length: int, overlap: int = 0):
+    """Batch data from iterable into slices of length N. The last batch may be shorter."""
+    # batched('ABCDEFG', 3) --> ABC DEF G
+    if max_batch_length < 1:
+        raise ValueError("n must be at least one")
+    for i in range(0, len(iterable), max_batch_length - overlap):
+        yield iterable[i : i + max_batch_length]
 
 
 def clean_input(prompt: str = "", talk=False):
@@ -49,7 +56,7 @@ def clean_input(prompt: str = "", talk=False):
 
         # ask for input, default when just pressing Enter is y
         logger.info("Asking user via keyboard...")
-        answer = input(prompt)
+        answer = session.prompt(ANSI(prompt))
         return answer
     except KeyboardInterrupt:
         logger.info("You interrupted Auto-GPT")
@@ -107,15 +114,69 @@ def get_current_git_branch() -> str:
         return ""
 
 
-def get_latest_bulletin() -> str:
-    exists = os.path.exists("CURRENT_BULLETIN.md")
+def get_latest_bulletin() -> tuple[str, bool]:
+    exists = os.path.exists("data/CURRENT_BULLETIN.md")
     current_bulletin = ""
     if exists:
-        current_bulletin = open("CURRENT_BULLETIN.md", "r", encoding="utf-8").read()
+        current_bulletin = open(
+            "data/CURRENT_BULLETIN.md", "r", encoding="utf-8"
+        ).read()
     new_bulletin = get_bulletin_from_web()
-    is_new_news = new_bulletin != current_bulletin
+    is_new_news = new_bulletin != "" and new_bulletin != current_bulletin
+
+    news_header = Fore.YELLOW + "Welcome to Auto-GPT!\n"
+    if new_bulletin or current_bulletin:
+        news_header += (
+            "Below you'll find the latest Auto-GPT News and updates regarding features!\n"
+            "If you don't wish to see this message, you "
+            "can run Auto-GPT with the *--skip-news* flag.\n"
+        )
 
     if new_bulletin and is_new_news:
-        open("CURRENT_BULLETIN.md", "w", encoding="utf-8").write(new_bulletin)
-        return f" {Fore.RED}::UPDATED:: {Fore.CYAN}{new_bulletin}{Fore.RESET}"
-    return current_bulletin
+        open("data/CURRENT_BULLETIN.md", "w", encoding="utf-8").write(new_bulletin)
+        current_bulletin = f"{Fore.RED}::NEW BULLETIN::{Fore.RESET}\n\n{new_bulletin}"
+
+    return f"{news_header}\n{current_bulletin}", is_new_news
+
+
+def markdown_to_ansi_style(markdown: str):
+    ansi_lines: list[str] = []
+    for line in markdown.split("\n"):
+        line_style = ""
+
+        if line.startswith("# "):
+            line_style += Style.BRIGHT
+        else:
+            line = re.sub(
+                r"(?<!\*)\*(\*?[^*]+\*?)\*(?!\*)",
+                rf"{Style.BRIGHT}\1{Style.NORMAL}",
+                line,
+            )
+
+        if re.match(r"^#+ ", line) is not None:
+            line_style += Fore.CYAN
+            line = re.sub(r"^#+ ", "", line)
+
+        ansi_lines.append(f"{line_style}{line}{Style.RESET_ALL}")
+    return "\n".join(ansi_lines)
+
+
+def get_legal_warning() -> str:
+    legal_text = """
+## DISCLAIMER AND INDEMNIFICATION AGREEMENT
+### PLEASE READ THIS DISCLAIMER AND INDEMNIFICATION AGREEMENT CAREFULLY BEFORE USING THE AUTOGPT SYSTEM. BY USING THE AUTOGPT SYSTEM, YOU AGREE TO BE BOUND BY THIS AGREEMENT.
+
+## Introduction
+AutoGPT (the "System") is a project that connects a GPT-like artificial intelligence system to the internet and allows it to automate tasks. While the System is designed to be useful and efficient, there may be instances where the System could perform actions that may cause harm or have unintended consequences.
+
+## No Liability for Actions of the System
+The developers, contributors, and maintainers of the AutoGPT project (collectively, the "Project Parties") make no warranties or representations, express or implied, about the System's performance, accuracy, reliability, or safety. By using the System, you understand and agree that the Project Parties shall not be liable for any actions taken by the System or any consequences resulting from such actions.
+
+## User Responsibility and Respondeat Superior Liability
+As a user of the System, you are responsible for supervising and monitoring the actions of the System while it is operating on your
+behalf. You acknowledge that using the System could expose you to potential liability including but not limited to respondeat superior and you agree to assume all risks and liabilities associated with such potential liability.
+
+## Indemnification
+By using the System, you agree to indemnify, defend, and hold harmless the Project Parties from and against any and all claims, liabilities, damages, losses, or expenses (including reasonable attorneys' fees and costs) arising out of or in connection with your use of the System, including, without limitation, any actions taken by the System on your behalf, any failure to properly supervise or monitor the System, and any resulting harm or unintended consequences.
+            """
+    return legal_text

autogpt/workspace/workspace.py

@@ -120,7 +120,8 @@ class Workspace:
 
         logger.debug(f"Resolved root as '{root}'")
 
-        if relative_path.is_absolute():
+        # Allow exception for absolute paths if they are contained in your workspace directory.
+        if relative_path.is_absolute() and not relative_path.is_relative_to(root):
             raise ValueError(
                 f"Attempted to access absolute path '{relative_path}' in workspace '{root}'."
             )

benchmark/benchmark_entrepreneur_gpt_with_difficult_user.py

@@ -75,14 +75,13 @@ Needs improvement.
 Not what I need."""
     # TODO: add questions above, to distract it even more.
 
-    command = f"{sys.executable} -m autogpt"
+    command = [sys.executable, "-m", "autogpt"]
 
     process = subprocess.Popen(
         command,
         stdin=subprocess.PIPE,
         stdout=subprocess.PIPE,
         stderr=subprocess.PIPE,
-        shell=True,
     )
 
     stdout_output, stderr_output = process.communicate(input_data.encode())

data_ingestion.py

@@ -1,37 +1,36 @@
 import argparse
 import logging
 
-from autogpt.commands.file_operations import ingest_file, search_files
+from autogpt.commands.file_operations import ingest_file, list_files
 from autogpt.config import Config
-from autogpt.memory import get_memory
+from autogpt.memory.vector import VectorMemory, get_memory
 
 cfg = Config()
 
 
 def configure_logging():
     logging.basicConfig(
-        filemode="a",
         format="%(asctime)s,%(msecs)d %(name)s %(levelname)s %(message)s",
         datefmt="%H:%M:%S",
         level=logging.DEBUG,
         handlers=[
-            logging.FileHandler(filename="log-ingestion.txt"),
+            logging.FileHandler(filename="log-ingestion.txt", mode="a"),
             logging.StreamHandler(),
         ],
     )
     return logging.getLogger("AutoGPT-Ingestion")
 
 
-def ingest_directory(directory, memory, args):
+def ingest_directory(directory: str, memory: VectorMemory, args):
     """
     Ingest all files in a directory by calling the ingest_file function for each file.
 
     :param directory: The directory containing the files to ingest
     :param memory: An object with an add() method to store the chunks in memory
     """
-    global logger
+    logger = logging.getLogger("AutoGPT-Ingestion")
     try:
-        files = search_files(directory)
+        files = list_files(directory)
         for file in files:
             ingest_file(file, memory, args.max_length, args.overlap)
     except Exception as e:
@@ -68,7 +67,6 @@ def main() -> None:
         help="The max_length of each chunk when ingesting files (default: 4000)",
         default=4000,
     )
-
     args = parser.parse_args()
 
     # Initialize memory