github/AutoGPT
1
1
Fork 0
mirror of https://github.com/Significant-Gravitas/AutoGPT.git synced 2026-02-01 10:24:56 -05:00
Code Issues Packages Projects Releases Wiki Activity

Resolved conflicts

Browse Source
This commit is contained in:
meta-fx
2023-04-13 22:47:21 -05:00
parent 570f76bd51 a3024ca80d
commit 3d783e08bc
53 changed files with 1866 additions and 433 deletions
Download Patch File Download Diff File Expand all files Collapse all files

0
scripts/__init__.py Normal file
View File

5
scripts/agent_manager.py
View File

@@ -6,6 +6,7 @@ agents = {} # key, (task, full_message_history, model)
# Create new GPT agent
# TODO: Centralise use of create_chat_completion() to globally enforce token limit
def create_agent(task, prompt, model):
"""Create a new agent and return its key"""
global next_key
@@ -13,7 +14,7 @@ def create_agent(task, prompt, model):
messages = [{"role": "user", "content": prompt}, ]
# Start GTP3 instance
# Start GPT instance
agent_reply = create_chat_completion(
model=model,
messages=messages,
@@ -41,7 +42,7 @@ def message_agent(key, message):
# Add user message to message history before sending to agent
messages.append({"role": "user", "content": message})
# Start GTP3 instance
# Start GPT instance
agent_reply = create_chat_completion(
model=model,
messages=messages,

11
scripts/ai_config.py
View File

@@ -1,6 +1,7 @@
import yaml
import data
import os
from prompt import get_prompt
class AIConfig:
"""
@@ -42,11 +43,11 @@ class AIConfig:
config_file (int): The path to the config yaml file. DEFAULT: "../ai_settings.yaml"
Returns:
cls (object): A instance of given cls object
cls (object): An instance of given cls object
"""
try:
with open(config_file) as file:
with open(config_file, encoding='utf-8') as file:
config_params = yaml.load(file, Loader=yaml.FullLoader)
except FileNotFoundError:
config_params = {}
@@ -80,7 +81,7 @@ class AIConfig:
None
Returns:
full_prompt (str): A string containing the intitial prompt for the user including the ai_name, ai_role and ai_goals.
full_prompt (str): A string containing the initial prompt for the user including the ai_name, ai_role and ai_goals.
"""
prompt_start = """Your decisions must always be made independently without seeking user assistance. Play to your strengths as an LLM and pursue simple strategies with no legal complications."""
@@ -90,5 +91,5 @@ class AIConfig:
for i, goal in enumerate(self.ai_goals):
full_prompt += f"{i+1}. {goal}\n"
full_prompt += f"\n\n{data.load_prompt()}"
full_prompt += f"\n\n{get_prompt()}"
return full_prompt

4
scripts/ai_functions.py
View File

@@ -1,8 +1,7 @@
from typing import List, Optional
from typing import List
import json
from config import Config
from call_ai_function import call_ai_function
from json_parser import fix_and_parse_json
cfg = Config()
@@ -46,7 +45,6 @@ def improve_code(suggestions: List[str], code: str) -> str:
return result_string
def write_tests(code: str, focus: List[str]) -> str:
"""
A function that takes in code and focus topics and returns a response from create chat completion api call.

73
scripts/browse.py
View File

@@ -2,32 +2,64 @@ import requests
from bs4 import BeautifulSoup
from config import Config
from llm_utils import create_chat_completion
from urllib.parse import urlparse, urljoin
cfg = Config()
# Function to check if the URL is valid
def is_valid_url(url):
try:
result = urlparse(url)
return all([result.scheme, result.netloc])
except ValueError:
return False
# Function to sanitize the URL
def sanitize_url(url):
return urljoin(url, urlparse(url).path)
# Define and check for local file address prefixes
def check_local_file_access(url):
local_prefixes = ['file:///', 'file://localhost', 'http://localhost', 'https://localhost']
return any(url.startswith(prefix) for prefix in local_prefixes)
def get_response(url, headers=cfg.user_agent_header, timeout=10):
try:
# Restrict access to local files
if check_local_file_access(url):
raise ValueError('Access to local files is restricted')
# Most basic check if the URL is valid:
if not url.startswith('http://') and not url.startswith('https://'):
raise ValueError('Invalid URL format')
sanitized_url = sanitize_url(url)
response = requests.get(sanitized_url, headers=headers, timeout=timeout)
# Check if the response contains an HTTP error
if response.status_code >= 400:
return None, "Error: HTTP " + str(response.status_code) + " error"
return response, None
except ValueError as ve:
# Handle invalid URL format
return None, "Error: " + str(ve)
except requests.exceptions.RequestException as re:
# Handle exceptions related to the HTTP request (e.g., connection errors, timeouts, etc.)
return None, "Error: " + str(re)
def scrape_text(url):
"""Scrape text from a webpage"""
# Most basic check if the URL is valid:
if not url.startswith('http'):
return "Error: Invalid URL"
# Restrict access to local files
if check_local_file_access(url):
return "Error: Access to local files is restricted"
try:
response = requests.get(url, headers=cfg.user_agent_header)
except requests.exceptions.RequestException as e:
return "Error: " + str(e)
# Check if the response contains an HTTP error
if response.status_code >= 400:
return "Error: HTTP " + str(response.status_code) + " error"
response, error_message = get_response(url)
if error_message:
return error_message
soup = BeautifulSoup(response.text, "html.parser")
@@ -60,11 +92,9 @@ def format_hyperlinks(hyperlinks):
def scrape_links(url):
"""Scrape links from a webpage"""
response = requests.get(url, headers=cfg.user_agent_header)
# Check if the response contains an HTTP error
if response.status_code >= 400:
return "error"
response, error_message = get_response(url)
if error_message:
return error_message
soup = BeautifulSoup(response.text, "html.parser")
@@ -102,6 +132,7 @@ def create_message(chunk, question):
"content": f"\"\"\"{chunk}\"\"\" Using the above text, please answer the following question: \"{question}\" -- if the question cannot be answered using the text, please summarize the text."
}
def summarize_text(text, question):
"""Summarize text using the LLM model"""
if not text:

2
scripts/call_ai_function.py
View File

@@ -3,6 +3,8 @@ from config import Config
cfg = Config()
from llm_utils import create_chat_completion
# This is a magic function that can do anything with no-code. See
# https://github.com/Torantulino/AI-Functions for more info.
def call_ai_function(function, args, description, model=None):

38
scripts/chat.py
View File

@@ -4,9 +4,12 @@ from dotenv import load_dotenv
from config import Config
import token_counter
from llm_utils import create_chat_completion
from logger import logger
import logging
cfg = Config()
def create_chat_message(role, content):
"""
Create a chat message with the given role and content.
@@ -64,15 +67,12 @@ def chat_with_ai(
model = cfg.fast_llm_model # TODO: Change model from hardcode to argument
# Reserve 1000 tokens for the response
if cfg.debug:
print(f"Token limit: {token_limit}")
logger.debug(f"Token limit: {token_limit}")
send_token_limit = token_limit - 1000
relevant_memory = permanent_memory.get_relevant(str(full_message_history[-5:]), 10)
relevant_memory = '' if len(full_message_history) ==0 else permanent_memory.get_relevant(str(full_message_history[-9:]), 10)
if cfg.debug:
print('Memory Stats: ', permanent_memory.get_stats())
logger.debug(f'Memory Stats: {permanent_memory.get_stats()}')
next_message_to_add_index, current_tokens_used, insertion_index, current_context = generate_context(
prompt, relevant_memory, full_message_history, model)
@@ -110,19 +110,17 @@ def chat_with_ai(
# 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
if cfg.debug:
print(f"Token limit: {token_limit}")
print(f"Send Token Count: {current_tokens_used}")
print(f"Tokens remaining for response: {tokens_remaining}")
print("------------ CONTEXT SENT TO AI ---------------")
for message in current_context:
# Skip printing the prompt
if message["role"] == "system" and message["content"] == prompt:
continue
print(
f"{message['role'].capitalize()}: {message['content']}")
print()
print("----------- END OF 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(
@@ -141,6 +139,6 @@ def chat_with_ai(
return assistant_reply
except openai.error.RateLimitError:
# TODO: WHen we switch to langchain, this is built in
# TODO: When we switch to langchain, this is built in
print("Error: ", "API Rate Limit Reached. Waiting 10 seconds...")
time.sleep(10)

12
scripts/commands.py
View File

@@ -7,7 +7,7 @@ import speak
from config import Config
import ai_functions as ai
from file_operations import read_file, write_to_file, append_to_file, delete_file, search_files
from execute_code import execute_python_file
from execute_code import execute_python_file, execute_shell
from json_parser import fix_and_parse_json
from image_gen import generate_image
from duckduckgo_search import ddg
@@ -24,6 +24,7 @@ def is_valid_int(value):
except ValueError:
return False
def get_command(response):
"""Parse the response and return the command name and arguments"""
try:
@@ -103,6 +104,11 @@ def execute_command(command_name, arguments):
return ai.write_tests(arguments["code"], arguments.get("focus"))
elif command_name == "execute_python_file": # Add this command
return execute_python_file(arguments["file"])
elif command_name == "execute_shell":
if cfg.execute_local_commands:
return execute_shell(arguments["command_line"])
else:
return "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."
elif command_name == "generate_image":
return generate_image(arguments["prompt"])
elif command_name == "do_nothing":
@@ -110,7 +116,7 @@ def execute_command(command_name, arguments):
elif command_name == "task_complete":
shutdown()
else:
return f"Unknown command '{command_name}'. Please refer to the 'COMMANDS' list for availabe commands and only respond in the specified JSON format."
return f"Unknown command '{command_name}'. Please refer to the 'COMMANDS' list for available commands and only respond in the specified JSON format."
# All errors, return "Error: + error message"
except Exception as e:
return "Error: " + str(e)
@@ -130,6 +136,7 @@ def google_search(query, num_results=8):
return json.dumps(search_results, ensure_ascii=False, indent=4)
def google_official_search(query, num_results=8):
"""Return the results of a google search using the official Google API"""
from googleapiclient.discovery import build
@@ -166,6 +173,7 @@ def google_official_search(query, num_results=8):
# Return the list of search result URLs
return search_results_links
def browse_website(url, question):
"""Browse a website and return the summary and links"""
summary = get_text_summary(url, question)

71
scripts/config.py
View File

@@ -1,6 +1,7 @@
import abc
import os
import openai
import yaml
from dotenv import load_dotenv
# Load environment variables from .env file
load_dotenv()
@@ -35,6 +36,7 @@ class Config(metaclass=Singleton):
"""Initialize the Config class"""
self.debug_mode = False
self.continuous_mode = False
self.continuous_limit = 0
self.speak_mode = False
self.fast_llm_model = os.getenv("FAST_LLM_MODEL", "gpt-3.5-turbo")
@@ -43,17 +45,19 @@ class Config(metaclass=Singleton):
self.smart_token_limit = int(os.getenv("SMART_TOKEN_LIMIT", 8000))
self.openai_api_key = os.getenv("OPENAI_API_KEY")
self.use_azure = False
self.temperature = float(os.getenv("TEMPERATURE", "1"))
self.use_azure = os.getenv("USE_AZURE") == 'True'
self.execute_local_commands = os.getenv('EXECUTE_LOCAL_COMMANDS', 'False') == 'True'
if self.use_azure:
self.openai_api_base = os.getenv("OPENAI_AZURE_API_BASE")
self.openai_api_version = os.getenv("OPENAI_AZURE_API_VERSION")
self.openai_deployment_id = os.getenv("OPENAI_AZURE_DEPLOYMENT_ID")
openai.api_type = "azure"
self.load_azure_config()
openai.api_type = self.openai_api_type
openai.api_base = self.openai_api_base
openai.api_version = self.openai_api_version
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")
self.use_mac_os_tts = False
self.use_mac_os_tts = os.getenv("USE_MAC_OS_TTS")
@@ -72,22 +76,67 @@ class Config(metaclass=Singleton):
# User agent headers to use when browsing web
# Some websites might just completely deny request with an error code if no user agent was found.
self.user_agent_header = {"User-Agent":"Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_4) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/83.0.4103.97 Safari/537.36"}
self.user_agent_header = {"User-Agent": "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_4) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/83.0.4103.97 Safari/537.36"}
self.redis_host = os.getenv("REDIS_HOST", "localhost")
self.redis_port = 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 configureable.
# Note that indexes must be created on db 0 in redis, this is not configurable.
self.memory_backend = os.getenv("MEMORY_BACKEND", 'local')
# Initialize the OpenAI API client
openai.api_key = self.openai_api_key
def get_azure_deployment_id_for_model(self, model: str) -> str:
"""
Returns the relevant deployment id for the model specified.
Parameters:
model(str): The model to map to the deployment id.
Returns:
The matching deployment id if found, otherwise an empty string.
"""
if model == self.fast_llm_model:
return self.azure_model_to_deployment_id_map["fast_llm_model_deployment_id"]
elif model == self.smart_llm_model:
return self.azure_model_to_deployment_id_map["smart_llm_model_deployment_id"]
elif model == "text-embedding-ada-002":
return self.azure_model_to_deployment_id_map["embedding_model_deployment_id"]
else:
return ""
AZURE_CONFIG_FILE = os.path.join(os.path.dirname(__file__), '..', 'azure.yaml')
def load_azure_config(self, config_file: str=AZURE_CONFIG_FILE) -> None:
"""
Loads the configuration parameters for Azure hosting from the specified file path as a yaml file.
Parameters:
config_file(str): The path to the config yaml file. DEFAULT: "../azure.yaml"
Returns:
None
"""
try:
with open(config_file) as file:
config_params = yaml.load(file, Loader=yaml.FullLoader)
except FileNotFoundError:
config_params = {}
self.openai_api_type = os.getenv("OPENAI_API_TYPE", config_params.get("azure_api_type", "azure"))
self.openai_api_base = os.getenv("OPENAI_AZURE_API_BASE", config_params.get("azure_api_base", ""))
self.openai_api_version = os.getenv("OPENAI_AZURE_API_VERSION", config_params.get("azure_api_version", ""))
self.azure_model_to_deployment_id_map = config_params.get("azure_model_map", [])
def set_continuous_mode(self, value: bool):
"""Set the continuous mode value."""
self.continuous_mode = value
def set_continuous_limit(self, value: int):
"""Set the continuous limit value."""
self.continuous_limit = value
def set_speak_mode(self, value: bool):
"""Set the speak mode value."""
self.speak_mode = value
@@ -116,6 +165,14 @@ class Config(metaclass=Singleton):
"""Set the ElevenLabs API key value."""
self.elevenlabs_api_key = value
def set_elevenlabs_voice_1_id(self, value: str):
"""Set the ElevenLabs Voice 1 ID value."""
self.elevenlabs_voice_1_id = value
def set_elevenlabs_voice_2_id(self, value: str):
"""Set the ElevenLabs Voice 2 ID value."""
self.elevenlabs_voice_2_id = value
def set_google_api_key(self, value: str):
"""Set the Google API key value."""
self.google_api_key = value

18
scripts/data.py
View File

@@ -1,18 +0,0 @@
import os
from pathlib import Path
def load_prompt():
"""Load the prompt from data/prompt.txt"""
try:
# get directory of this file:
file_dir = Path(__file__).parent
prompt_file_path = file_dir / "data" / "prompt.txt"
# Load the prompt from data/prompt.txt
with open(prompt_file_path, "r") as prompt_file:
prompt = prompt_file.read()
return prompt
except FileNotFoundError:
print("Error: Prompt file not found", flush=True)
return ""

63
scripts/data/prompt.txt
View File

@@ -1,63 +0,0 @@
CONSTRAINTS:
1. ~4000 word limit for short term memory. Your short term memory is short, so immediately save important information to files.
2. If you are unsure how you previously did something or want to recall past events, thinking about similar events will help you remember.
3. No user assistance
4. Exclusively use the commands listed in double quotes e.g. "command name"
COMMANDS:
1. Google Search: "google", args: "input": "<search>"
5. Browse Website: "browse_website", args: "url": "<url>", "question": "<what_you_want_to_find_on_website>"
6. Start GPT Agent: "start_agent", args: "name": "<name>", "task": "<short_task_desc>", "prompt": "<prompt>"
7. Message GPT Agent: "message_agent", args: "key": "<key>", "message": "<message>"
8. List GPT Agents: "list_agents", args: ""
9. Delete GPT Agent: "delete_agent", args: "key": "<key>"
10. Write to file: "write_to_file", args: "file": "<file>", "text": "<text>"
11. Read file: "read_file", args: "file": "<file>"
12. Append to file: "append_to_file", args: "file": "<file>", "text": "<text>"
13. Delete file: "delete_file", args: "file": "<file>"
14. Search Files: "search_files", args: "directory": "<directory>"
15. Evaluate Code: "evaluate_code", args: "code": "<full_code_string>"
16. Get Improved Code: "improve_code", args: "suggestions": "<list_of_suggestions>", "code": "<full_code_string>"
17. Write Tests: "write_tests", args: "code": "<full_code_string>", "focus": "<list_of_focus_areas>"
18. Execute Python File: "execute_python_file", args: "file": "<file>"
19. Task Complete (Shutdown): "task_complete", args: "reason": "<reason>"
20. Generate Image: "generate_image", args: "prompt": "<prompt>"
21. Do Nothing: "do_nothing", args: ""
RESOURCES:
1. Internet access for searches and information gathering.
2. Long Term memory management.
3. GPT-3.5 powered Agents for delegation of simple tasks.
4. File output.
PERFORMANCE EVALUATION:
1. Continuously review and analyze your actions to ensure you are performing to the best of your abilities.
2. Constructively self-criticize your big-picture behavior constantly.
3. Reflect on past decisions and strategies to refine your approach.
4. Every command has a cost, so be smart and efficient. Aim to complete tasks in the least number of steps.
You should only respond in JSON format as described below
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"
}
}
}
Ensure the response can be parsed by Python json.loads

50
scripts/execute_code.py
View File

@@ -1,17 +1,20 @@
import docker
import os
import subprocess
WORKSPACE_FOLDER = "auto_gpt_workspace"
def execute_python_file(file):
"""Execute a Python file in a Docker container and return the output"""
workspace_folder = "auto_gpt_workspace"
print (f"Executing file '{file}' in workspace '{workspace_folder}'")
print (f"Executing file '{file}' in workspace '{WORKSPACE_FOLDER}'")
if not file.endswith(".py"):
return "Error: Invalid file type. Only .py files are allowed."
file_path = os.path.join(workspace_folder, file)
file_path = os.path.join(WORKSPACE_FOLDER, file)
if not os.path.isfile(file_path):
return f"Error: File '{file}' does not exist."
@@ -19,14 +22,31 @@ def execute_python_file(file):
try:
client = docker.from_env()
image_name = 'python:3.10'
try:
client.images.get(image_name)
print(f"Image '{image_name}' found locally")
except docker.errors.ImageNotFound:
print(f"Image '{image_name}' not found locally, pulling from Docker Hub")
# Use the low-level API to stream the pull response
low_level_client = docker.APIClient()
for line in low_level_client.pull(image_name, stream=True, decode=True):
# Print the status and progress, if available
status = line.get('status')
progress = line.get('progress')
if status and progress:
print(f"{status}: {progress}")
elif status:
print(status)
# You can replace 'python:3.8' with the desired Python image/version
# You can find available Python images on Docker Hub:
# https://hub.docker.com/_/python
container = client.containers.run(
'python:3.10',
image_name,
f'python {file}',
volumes={
os.path.abspath(workspace_folder): {
os.path.abspath(WORKSPACE_FOLDER): {
'bind': '/workspace',
'mode': 'ro'}},
working_dir='/workspace',
@@ -46,3 +66,23 @@ def execute_python_file(file):
except Exception as e:
return f"Error: {str(e)}"
def execute_shell(command_line):
current_dir = os.getcwd()
if not WORKSPACE_FOLDER in current_dir: # Change dir into workspace if necessary
work_dir = os.path.join(os.getcwd(), WORKSPACE_FOLDER)
os.chdir(work_dir)
print (f"Executing command '{command_line}' in working directory '{os.getcwd()}'")
result = subprocess.run(command_line, capture_output=True, shell=True)
output = f"STDOUT:\n{result.stdout}\nSTDERR:\n{result.stderr}"
# Change back to whatever the prior working dir was
os.chdir(current_dir)
return output

5
scripts/file_operations.py
View File

@@ -38,7 +38,7 @@ def write_to_file(filename, text):
directory = os.path.dirname(filepath)
if not os.path.exists(directory):
os.makedirs(directory)
with open(filepath, "w") as f:
with open(filepath, "w", encoding='utf-8') as f:
f.write(text)
return "File written to successfully."
except Exception as e:
@@ -65,6 +65,7 @@ def delete_file(filename):
except Exception as e:
return "Error: " + str(e)
def search_files(directory):
found_files = []
@@ -80,4 +81,4 @@ def search_files(directory):
relative_path = os.path.relpath(os.path.join(root, file), working_directory)
found_files.append(relative_path)
return found_files
return found_files

1
scripts/image_gen.py
View File

@@ -11,6 +11,7 @@ cfg = Config()
working_directory = "auto_gpt_workspace"
def generate_image(prompt):
filename = str(uuid.uuid4()) + ".jpg"

32
scripts/json_parser.py
View File

@@ -3,6 +3,7 @@ from typing import Any, Dict, Union
from call_ai_function import call_ai_function
from config import Config
from json_utils import correct_json
from logger import logger
cfg = Config()
@@ -26,7 +27,7 @@ JSON_SCHEMA = """
"""
def fix_and_parse_json(
def fix_and_parse_json(
json_str: str,
try_to_fix_with_gpt: bool = True
) -> Union[str, Dict[Any, Any]]:
@@ -35,8 +36,8 @@ def fix_and_parse_json(
json_str = json_str.replace('\t', '')
return json.loads(json_str)
except json.JSONDecodeError as _: # noqa: F841
json_str = correct_json(json_str)
try:
json_str = correct_json(json_str)
return json.loads(json_str)
except json.JSONDecodeError as _: # noqa: F841
pass
@@ -53,9 +54,10 @@ def fix_and_parse_json(
last_brace_index = json_str.rindex("}")
json_str = json_str[:last_brace_index+1]
return json.loads(json_str)
except json.JSONDecodeError as e: # noqa: F841
# Can throw a ValueError if there is no "{" or "}" in the json_str
except (json.JSONDecodeError, ValueError) as e: # noqa: F841
if try_to_fix_with_gpt:
print("Warning: Failed to parse AI output, attempting to fix."
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.")
@@ -67,22 +69,21 @@ def fix_and_parse_json(
else:
# This allows the AI to react to the error message,
# which usually results in it correcting its ways.
print("Failed to fix ai output, telling the AI.")
logger.error("Failed to fix AI output, telling the AI.")
return json_str
else:
raise e
def fix_json(json_str: str, schema: str) -> str:
"""Fix the given JSON string to make it parseable and fully complient with the provided schema."""
# Try to fix the JSON using gpt:
"""Fix the given JSON string to make it parseable and fully compliant with the provided schema."""
# Try to fix the JSON using GPT:
function_string = "def fix_json(json_str: str, schema:str=None) -> str:"
args = [f"'''{json_str}'''", f"'''{schema}'''"]
description_string = "Fixes the provided JSON string to make it parseable"\
" and fully complient with the provided schema.\n If an object or"\
" and fully compliant with the provided schema.\n If an object or"\
" field specified in the schema isn't contained within the correct"\
" JSON, it is ommited.\n This function is brilliant at guessing"\
" JSON, it is omitted.\n This function is brilliant at guessing"\
" when the format is incorrect."
# If it doesn't already start with a "`", add one:
@@ -91,12 +92,11 @@ def fix_json(json_str: str, schema: str) -> str:
result_string = call_ai_function(
function_string, args, description_string, model=cfg.fast_llm_model
)
if cfg.debug:
print("------------ JSON FIX ATTEMPT ---------------")
print(f"Original JSON: {json_str}")
print("-----------")
print(f"Fixed JSON: {result_string}")
print("----------- END OF FIX ATTEMPT ----------------")
logger.debug("------------ JSON FIX ATTEMPT ---------------")
logger.debug(f"Original JSON: {json_str}")
logger.debug("-----------")
logger.debug(f"Fixed JSON: {result_string}")
logger.debug("----------- END OF FIX ATTEMPT ----------------")
try:
json.loads(result_string) # just check the validity

58
scripts/llm_utils.py
View File

@@ -1,26 +1,52 @@
import time
import openai
from colorama import Fore
from config import Config
cfg = Config()
openai.api_key = cfg.openai_api_key
# Overly simple abstraction until we create something better
def create_chat_completion(messages, model=None, temperature=None, max_tokens=None)->str:
# simple retry mechanism when getting a rate error or a bad gateway
def create_chat_completion(messages, model=None, temperature=cfg.temperature, max_tokens=None)->str:
"""Create a chat completion using the OpenAI API"""
if cfg.use_azure:
response = openai.ChatCompletion.create(
deployment_id=cfg.openai_deployment_id,
model=model,
messages=messages,
temperature=temperature,
max_tokens=max_tokens
)
else:
response = openai.ChatCompletion.create(
model=model,
messages=messages,
temperature=temperature,
max_tokens=max_tokens
)
response = None
num_retries = 5
for attempt in range(num_retries):
try:
if cfg.use_azure:
response = openai.ChatCompletion.create(
deployment_id=cfg.get_azure_deployment_id_for_model(model),
model=model,
messages=messages,
temperature=temperature,
max_tokens=max_tokens
)
else:
response = openai.ChatCompletion.create(
model=model,
messages=messages,
temperature=temperature,
max_tokens=max_tokens
)
break
except openai.error.RateLimitError:
if cfg.debug_mode:
print(Fore.RED + "Error: ", "API Rate Limit Reached. Waiting 20 seconds..." + Fore.RESET)
time.sleep(20)
except openai.error.APIError as e:
if e.http_status == 502:
if cfg.debug_mode:
print(Fore.RED + "Error: ", "API Bad gateway. Waiting 20 seconds..." + Fore.RESET)
time.sleep(20)
else:
raise
if attempt == num_retries - 1:
raise
if response is None:
raise RuntimeError("Failed to get response after 5 retries")
return response.choices[0].message["content"]

192
scripts/logger.py Normal file
View File

@@ -0,0 +1,192 @@
import logging
import os
import random
import re
import time
from logging import LogRecord
from colorama import Fore
from colorama import Style
import speak
from config import Config
from config import Singleton
cfg = Config()
'''
Logger that handle titles in different colors.
Outputs logs in console, activity.log, and errors.log
For console handler: simulates typing
'''
class Logger(metaclass=Singleton):
def __init__(self):
# create log directory if it doesn't exist
log_dir = os.path.join('..', 'logs')
if not os.path.exists(log_dir):
os.makedirs(log_dir)
log_file = "activity.log"
error_file = "error.log"
console_formatter = AutoGptFormatter('%(title_color)s %(message)s')
# Create a handler for console which simulate typing
self.typing_console_handler = TypingConsoleHandler()
self.typing_console_handler.setLevel(logging.INFO)
self.typing_console_handler.setFormatter(console_formatter)
# Create a handler for console without typing simulation
self.console_handler = ConsoleHandler()
self.console_handler.setLevel(logging.DEBUG)
self.console_handler.setFormatter(console_formatter)
# Info handler in activity.log
self.file_handler = logging.FileHandler(os.path.join(log_dir, log_file))
self.file_handler.setLevel(logging.DEBUG)
info_formatter = AutoGptFormatter('%(asctime)s %(levelname)s %(title)s %(message_no_color)s')
self.file_handler.setFormatter(info_formatter)
# Error handler error.log
error_handler = logging.FileHandler(os.path.join(log_dir, error_file))
error_handler.setLevel(logging.ERROR)
error_formatter = AutoGptFormatter(
'%(asctime)s %(levelname)s %(module)s:%(funcName)s:%(lineno)d %(title)s %(message_no_color)s')
error_handler.setFormatter(error_formatter)
self.typing_logger = logging.getLogger('TYPER')
self.typing_logger.addHandler(self.typing_console_handler)
self.typing_logger.addHandler(self.file_handler)
self.typing_logger.addHandler(error_handler)
self.typing_logger.setLevel(logging.DEBUG)
self.logger = logging.getLogger('LOGGER')
self.logger.addHandler(self.console_handler)
self.logger.addHandler(self.file_handler)
self.logger.addHandler(error_handler)
self.logger.setLevel(logging.DEBUG)
def typewriter_log(
self,
title='',
title_color='',
content='',
speak_text=False,
level=logging.INFO):
if speak_text and cfg.speak_mode:
speak.say_text(f"{title}. {content}")
if content:
if isinstance(content, list):
content = " ".join(content)
else:
content = ""
self.typing_logger.log(level, content, extra={'title': title, 'color': title_color})
def debug(
self,
message,
title='',
title_color='',
):
self._log(title, title_color, message, logging.DEBUG)
def warn(
self,
message,
title='',
title_color='',
):
self._log(title, title_color, message, logging.WARN)
def error(
self,
title,
message=''
):
self._log(title, Fore.RED, message, logging.ERROR)
def _log(
self,
title='',
title_color='',
message='',
level=logging.INFO):
if message:
if isinstance(message, list):
message = " ".join(message)
self.logger.log(level, message, extra={'title': title, 'color': title_color})
def set_level(self, level):
self.logger.setLevel(level)
self.typing_logger.setLevel(level)
def double_check(self, additionalText=None):
if not additionalText:
additionalText = "Please ensure you've setup and configured everything correctly. Read https://github.com/Torantulino/Auto-GPT#readme to double check. You can also create a github issue or join the discord and ask there!"
self.typewriter_log("DOUBLE CHECK CONFIGURATION", Fore.YELLOW, additionalText)
'''
Output stream to console using simulated typing
'''
class TypingConsoleHandler(logging.StreamHandler):
def emit(self, record):
min_typing_speed = 0.05
max_typing_speed = 0.01
msg = self.format(record)
try:
words = msg.split()
for i, word in enumerate(words):
print(word, end="", flush=True)
if i < len(words) - 1:
print(" ", end="", flush=True)
typing_speed = random.uniform(min_typing_speed, max_typing_speed)
time.sleep(typing_speed)
# type faster after each word
min_typing_speed = min_typing_speed * 0.95
max_typing_speed = max_typing_speed * 0.95
print()
except Exception:
self.handleError(record)
class ConsoleHandler(logging.StreamHandler):
def emit(self, record):
msg = self.format(record)
try:
print(msg)
except Exception:
self.handleError(record)
class AutoGptFormatter(logging.Formatter):
"""
Allows to handle custom placeholders 'title_color' and 'message_no_color'.
To use this formatter, make sure to pass 'color', 'title' as log extras.
"""
def format(self, record: LogRecord) -> str:
if (hasattr(record, 'color')):
record.title_color = getattr(record, 'color') + getattr(record, 'title') + " " + Style.RESET_ALL
else:
record.title_color = getattr(record, 'title')
if hasattr(record, 'msg'):
record.message_no_color = remove_color_codes(getattr(record, 'msg'))
else:
record.message_no_color = ''
return super().format(record)
def remove_color_codes(s: str) -> str:
ansi_escape = re.compile(r'\x1B(?:[@-Z\\-_]|\[[0-?]*[ -/]*[@-~])')
return ansi_escape.sub('', s)
logger = Logger()

393
scripts/main.py
View File

@@ -2,8 +2,7 @@ import json
import random
import commands as cmd
import utils
from memory import get_memory
import data
from memory import get_memory, get_supported_memory_backends
import chat
from colorama import Fore, Style
from spinner import Spinner
@@ -15,56 +14,51 @@ from ai_config import AIConfig
import traceback
import yaml
import argparse
from logger import logger
import logging
from prompt import get_prompt
cfg = Config()
def configure_logging():
logging.basicConfig(filename='log.txt',
filemode='a',
format='%(asctime)s,%(msecs)d %(name)s %(levelname)s %(message)s',
datefmt='%H:%M:%S',
level=logging.DEBUG)
return logging.getLogger('AutoGPT')
def check_openai_api_key():
"""Check if the OpenAI API key is set in config.py or as an environment variable."""
if not cfg.openai_api_key:
print(
Fore.RED +
"Please set your OpenAI API key in config.py or as an environment variable."
"Please set your OpenAI API key in .env or as an environment variable."
)
print("You can get your key from https://beta.openai.com/account/api-keys")
exit(1)
def print_to_console(
title,
title_color,
content,
speak_text=False,
min_typing_speed=0.05,
max_typing_speed=0.01):
"""Prints text to the console with a typing effect"""
global cfg
global logger
if speak_text and cfg.speak_mode:
speak.say_text(f"{title}. {content}")
print(title_color + title + " " + Style.RESET_ALL, end="")
if content:
logger.info(title + ': ' + content)
if isinstance(content, list):
content = " ".join(content)
words = content.split()
for i, word in enumerate(words):
print(word, end="", flush=True)
if i < len(words) - 1:
print(" ", end="", flush=True)
typing_speed = random.uniform(min_typing_speed, max_typing_speed)
time.sleep(typing_speed)
# type faster after each word
min_typing_speed = min_typing_speed * 0.95
max_typing_speed = max_typing_speed * 0.95
print()
def attempt_to_fix_json_by_finding_outermost_brackets(json_string):
if cfg.speak_mode and cfg.debug_mode:
speak.say_text("I have received an invalid JSON response from the OpenAI API. Trying to fix it now.")
logger.typewriter_log("Attempting to fix JSON by finding outermost brackets\n")
try:
# Use regex to search for JSON objects
import regex
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:
speak.say_text("Apparently json was fixed.")
else:
raise ValueError("No valid JSON object found")
except (json.JSONDecodeError, ValueError) as e:
if cfg.speak_mode:
speak.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 json_string
def print_assistant_thoughts(assistant_reply):
@@ -72,16 +66,21 @@ def print_assistant_thoughts(assistant_reply):
global ai_name
global cfg
try:
# Parse and print Assistant response
assistant_reply_json = fix_and_parse_json(assistant_reply)
try:
# Parse and print Assistant response
assistant_reply_json = fix_and_parse_json(assistant_reply)
except json.JSONDecodeError as e:
logger.error("Error: Invalid JSON in assistant thoughts\n", assistant_reply)
assistant_reply_json = attempt_to_fix_json_by_finding_outermost_brackets(assistant_reply)
assistant_reply_json = fix_and_parse_json(assistant_reply_json)
# Check if assistant_reply_json is a string and attempt to parse it into a JSON object
if isinstance(assistant_reply_json, str):
try:
assistant_reply_json = json.loads(assistant_reply_json)
except json.JSONDecodeError as e:
print_to_console("Error: Invalid JSON\n", Fore.RED, assistant_reply)
assistant_reply_json = {}
logger.error("Error: Invalid JSON\n", assistant_reply)
assistant_reply_json = attempt_to_fix_json_by_finding_outermost_brackets(assistant_reply_json)
assistant_thoughts_reasoning = None
assistant_thoughts_plan = None
@@ -96,11 +95,11 @@ def print_assistant_thoughts(assistant_reply):
assistant_thoughts_criticism = assistant_thoughts.get("criticism")
assistant_thoughts_speak = assistant_thoughts.get("speak")
print_to_console(f"{ai_name.upper()} THOUGHTS:", Fore.YELLOW, assistant_thoughts_text)
print_to_console("REASONING:", Fore.YELLOW, assistant_thoughts_reasoning)
logger.typewriter_log(f"{ai_name.upper()} THOUGHTS:", Fore.YELLOW, assistant_thoughts_text)
logger.typewriter_log("REASONING:", Fore.YELLOW, assistant_thoughts_reasoning)
if assistant_thoughts_plan:
print_to_console("PLAN:", Fore.YELLOW, "")
logger.typewriter_log("PLAN:", Fore.YELLOW, "")
# If it's a list, join it into a string
if isinstance(assistant_thoughts_plan, list):
assistant_thoughts_plan = "\n".join(assistant_thoughts_plan)
@@ -111,20 +110,23 @@ def print_assistant_thoughts(assistant_reply):
lines = assistant_thoughts_plan.split('\n')
for line in lines:
line = line.lstrip("- ")
print_to_console("- ", Fore.GREEN, line.strip())
logger.typewriter_log("- ", Fore.GREEN, line.strip())
print_to_console("CRITICISM:", Fore.YELLOW, assistant_thoughts_criticism)
logger.typewriter_log("CRITICISM:", Fore.YELLOW, assistant_thoughts_criticism)
# Speak the assistant's thoughts
if cfg.speak_mode and assistant_thoughts_speak:
speak.say_text(assistant_thoughts_speak)
except json.decoder.JSONDecodeError:
print_to_console("Error: Invalid JSON\n", Fore.RED, assistant_reply)
return assistant_reply_json
except json.decoder.JSONDecodeError as e:
logger.error("Error: Invalid JSON\n", assistant_reply)
if cfg.speak_mode:
speak.say_text("I have received an invalid JSON response from the OpenAI API. I cannot ignore this response.")
# All other errors, return "Error: + error message"
except Exception as e:
call_stack = traceback.format_exc()
print_to_console("Error: \n", Fore.RED, call_stack)
logger.error("Error: \n", call_stack)
def load_variables(config_file="config.yaml"):
@@ -169,8 +171,8 @@ def load_variables(config_file="config.yaml"):
with open(config_file, "w") as file:
documents = yaml.dump(config, file)
prompt = data.load_prompt()
prompt_start = """Your decisions must always be made independently without seeking user assistance. Play to your strengths as a LLM and pursue simple strategies with no legal complications."""
prompt = get_prompt()
prompt_start = """Your decisions must always be made independently without seeking user assistance. Play to your strengths as an LLM and pursue simple strategies with no legal complications."""
# Construct full prompt
full_prompt = f"You are {ai_name}, {ai_role}\n{prompt_start}\n\nGOALS:\n\n"
@@ -185,7 +187,7 @@ def construct_prompt():
"""Construct the prompt for the AI to respond to"""
config = AIConfig.load()
if config.ai_name:
print_to_console(
logger.typewriter_log(
f"Welcome back! ",
Fore.GREEN,
f"Would you like me to return to being {config.ai_name}?",
@@ -214,14 +216,14 @@ def prompt_user():
"""Prompt the user for input"""
ai_name = ""
# Construct the prompt
print_to_console(
logger.typewriter_log(
"Welcome to Auto-GPT! ",
Fore.GREEN,
"Enter the name of your AI and its role below. Entering nothing will load defaults.",
speak_text=True)
# Get AI Name from User
print_to_console(
logger.typewriter_log(
"Name your AI: ",
Fore.GREEN,
"For example, 'Entrepreneur-GPT'")
@@ -229,14 +231,14 @@ def prompt_user():
if ai_name == "":
ai_name = "Entrepreneur-GPT"
print_to_console(
logger.typewriter_log(
f"{ai_name} here!",
Fore.LIGHTBLUE_EX,
"I am at your service.",
speak_text=True)
# Get AI Role from User
print_to_console(
logger.typewriter_log(
"Describe your AI's role: ",
Fore.GREEN,
"For example, 'an AI designed to autonomously develop and run businesses with the sole goal of increasing your net worth.'")
@@ -245,7 +247,7 @@ def prompt_user():
ai_role = "an AI designed to autonomously develop and run businesses with the sole goal of increasing your net worth."
# Enter up to 5 goals for the AI
print_to_console(
logger.typewriter_log(
"Enter up to 5 goals for your AI: ",
Fore.GREEN,
"For example: \nIncrease net worth, Grow Twitter Account, Develop and manage multiple businesses autonomously'")
@@ -263,148 +265,197 @@ def prompt_user():
config = AIConfig(ai_name, ai_role, ai_goals)
return config
def parse_arguments():
"""Parses the arguments passed to the script"""
global cfg
cfg.set_debug_mode(False)
cfg.set_continuous_mode(False)
cfg.set_speak_mode(False)
parser = argparse.ArgumentParser(description='Process arguments.')
parser.add_argument('--continuous', action='store_true', help='Enable Continuous Mode')
parser.add_argument('--continuous-limit', '-l', type=int, dest="continuous_limit", help='Defines the number of times to run in continuous mode')
parser.add_argument('--speak', action='store_true', help='Enable Speak Mode')
parser.add_argument('--debug', action='store_true', help='Enable Debug Mode')
parser.add_argument('--gpt3only', action='store_true', help='Enable GPT3.5 Only Mode')
parser.add_argument('--gpt4only', action='store_true', help='Enable GPT4 Only Mode')
parser.add_argument('--use-memory', '-m', dest="memory_type", help='Defines which Memory backend to use')
args = parser.parse_args()
if args.debug:
logger.typewriter_log("Debug Mode: ", Fore.GREEN, "ENABLED")
cfg.set_debug_mode(True)
if args.continuous:
print_to_console("Continuous Mode: ", Fore.RED, "ENABLED")
print_to_console(
logger.typewriter_log("Continuous Mode: ", Fore.RED, "ENABLED")
logger.typewriter_log(
"WARNING: ",
Fore.RED,
"Continuous mode is not recommended. It is potentially dangerous and may 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)
if args.continuous_limit:
logger.typewriter_log(
"Continuous Limit: ",
Fore.GREEN,
f"{args.continuous_limit}")
cfg.set_continuous_limit(args.continuous_limit)
# Check if continuous limit is used without continuous mode
if args.continuous_limit and not args.continuous:
parser.error("--continuous-limit can only be used with --continuous")
if args.speak:
print_to_console("Speak Mode: ", Fore.GREEN, "ENABLED")
logger.typewriter_log("Speak Mode: ", Fore.GREEN, "ENABLED")
cfg.set_speak_mode(True)
if args.gpt3only:
print_to_console("GPT3.5 Only Mode: ", Fore.GREEN, "ENABLED")
logger.typewriter_log("GPT3.5 Only Mode: ", Fore.GREEN, "ENABLED")
cfg.set_smart_llm_model(cfg.fast_llm_model)
if args.gpt4only:
logger.typewriter_log("GPT4 Only Mode: ", Fore.GREEN, "ENABLED")
cfg.set_fast_llm_model(cfg.smart_llm_model)
if args.debug:
logger.typewriter_log("Debug Mode: ", Fore.GREEN, "ENABLED")
cfg.set_debug_mode(True)
if args.memory_type:
supported_memory = get_supported_memory_backends()
chosen = args.memory_type
if not chosen in supported_memory:
logger.typewriter_log("ONLY THE FOLLOWING MEMORY BACKENDS ARE SUPPORTED: ", Fore.RED, f'{supported_memory}')
logger.typewriter_log(f"Defaulting to: ", Fore.YELLOW, cfg.memory_backend)
else:
cfg.memory_backend = chosen
# TODO: fill in llm values here
check_openai_api_key()
cfg = Config()
logger = configure_logging()
parse_arguments()
ai_name = ""
prompt = construct_prompt()
# print(prompt)
# Initialize variables
full_message_history = []
result = None
next_action_count = 0
# Make a constant:
user_input = "Determine which next command to use, and respond using the format specified above:"
# Initialize memory and make sure it is empty.
# this is particularly important for indexing and referencing pinecone memory
memory = get_memory(cfg, init=True)
print('Using memory of type: ' + memory.__class__.__name__)
# Interaction Loop
while True:
# Send message to AI, get response
with Spinner("Thinking... "):
assistant_reply = chat.chat_with_ai(
prompt,
user_input,
full_message_history,
memory,
cfg.fast_token_limit) # TODO: This hardcodes the model to use GPT3.5. Make this an argument
# Print Assistant thoughts
print_assistant_thoughts(assistant_reply)
# Get command name and arguments
try:
command_name, arguments = cmd.get_command(assistant_reply)
except Exception as e:
print_to_console("Error: \n", Fore.RED, str(e))
if not cfg.continuous_mode and 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
user_input = ""
print_to_console(
"NEXT ACTION: ",
Fore.CYAN,
f"COMMAND = {Fore.CYAN}{command_name}{Style.RESET_ALL} ARGUMENTS = {Fore.CYAN}{arguments}{Style.RESET_ALL}")
print(
f"Enter 'y' to authorise command, 'y -N' to run N continuous commands, 'n' to exit program, or enter feedback for {ai_name}...",
flush=True)
while True:
console_input = utils.clean_input(Fore.MAGENTA + "Input:" + Style.RESET_ALL)
if console_input.lower() == "y":
user_input = "GENERATE NEXT COMMAND JSON"
break
elif console_input.lower().startswith("y -"):
try:
next_action_count = abs(int(console_input.split(" ")[1]))
user_input = "GENERATE NEXT COMMAND JSON"
except ValueError:
print("Invalid input format. Please enter 'y -n' where n is the number of continuous tasks.")
continue
break
elif console_input.lower() == "n":
user_input = "EXIT"
break
else:
user_input = console_input
command_name = "human_feedback"
break
if user_input == "GENERATE NEXT COMMAND JSON":
print_to_console(
"-=-=-=-=-=-=-= COMMAND AUTHORISED BY USER -=-=-=-=-=-=-=",
Fore.MAGENTA,
"")
elif user_input == "EXIT":
print("Exiting...", flush=True)
def main():
global ai_name, memory
# TODO: fill in llm values here
check_openai_api_key()
parse_arguments()
logger.set_level(logging.DEBUG if cfg.debug_mode else logging.INFO)
ai_name = ""
prompt = construct_prompt()
# print(prompt)
# Initialize variables
full_message_history = []
result = None
next_action_count = 0
# Make a constant:
user_input = "Determine which next command to use, and respond using the format specified above:"
# Initialize memory and make sure it is empty.
# this is particularly important for indexing and referencing pinecone memory
memory = get_memory(cfg, init=True)
print('Using memory of type: ' + memory.__class__.__name__)
# Interaction Loop
loop_count = 0
while True:
# Discontinue if continuous limit is reached
loop_count += 1
if cfg.continuous_mode and cfg.continuous_limit > 0 and loop_count > cfg.continuous_limit:
logger.typewriter_log("Continuous Limit Reached: ", Fore.YELLOW, f"{cfg.continuous_limit}")
break
else:
# Print command
print_to_console(
"NEXT ACTION: ",
Fore.CYAN,
f"COMMAND = {Fore.CYAN}{command_name}{Style.RESET_ALL} ARGUMENTS = {Fore.CYAN}{arguments}{Style.RESET_ALL}")
# Execute command
if command_name.lower().startswith( "error" ):
result = f"Command {command_name} threw the following error: " + arguments
elif command_name == "human_feedback":
result = f"Human feedback: {user_input}"
else:
result = f"Command {command_name} returned: {cmd.execute_command(command_name, arguments)}"
if next_action_count > 0:
next_action_count -= 1
# Send message to AI, get response
with Spinner("Thinking... "):
assistant_reply = chat.chat_with_ai(
prompt,
user_input,
full_message_history,
memory,
cfg.fast_token_limit) # TODO: This hardcodes the model to use GPT3.5. Make this an argument
memory_to_add = f"Assistant Reply: {assistant_reply} " \
f"\nResult: {result} " \
f"\nHuman Feedback: {user_input} "
# Print Assistant thoughts
print_assistant_thoughts(assistant_reply)
memory.add(memory_to_add)
# Get command name and arguments
try:
command_name, arguments = cmd.get_command(
attempt_to_fix_json_by_finding_outermost_brackets(assistant_reply))
if cfg.speak_mode:
speak.say_text(f"I want to execute {command_name}")
except Exception as e:
logger.error("Error: \n", str(e))
# Check if there's a result from the command append it to the message
# history
if result is not None:
full_message_history.append(chat.create_chat_message("system", result))
print_to_console("SYSTEM: ", Fore.YELLOW, result)
else:
full_message_history.append(
chat.create_chat_message(
"system", "Unable to execute command"))
print_to_console("SYSTEM: ", Fore.YELLOW, "Unable to execute command")
if not cfg.continuous_mode and 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
user_input = ""
logger.typewriter_log(
"NEXT ACTION: ",
Fore.CYAN,
f"COMMAND = {Fore.CYAN}{command_name}{Style.RESET_ALL} ARGUMENTS = {Fore.CYAN}{arguments}{Style.RESET_ALL}")
print(
f"Enter 'y' to authorise command, 'y -N' to run N continuous commands, 'n' to exit program, or enter feedback for {ai_name}...",
flush=True)
while True:
console_input = utils.clean_input(Fore.MAGENTA + "Input:" + Style.RESET_ALL)
if console_input.lower().rstrip() == "y":
user_input = "GENERATE NEXT COMMAND JSON"
break
elif console_input.lower().startswith("y -"):
try:
next_action_count = abs(int(console_input.split(" ")[1]))
user_input = "GENERATE NEXT COMMAND JSON"
except ValueError:
print("Invalid input format. Please enter 'y -n' where n is the number of continuous tasks.")
continue
break
elif console_input.lower() == "n":
user_input = "EXIT"
break
else:
user_input = console_input
command_name = "human_feedback"
break
if user_input == "GENERATE NEXT COMMAND JSON":
logger.typewriter_log(
"-=-=-=-=-=-=-= COMMAND AUTHORISED BY USER -=-=-=-=-=-=-=",
Fore.MAGENTA,
"")
elif user_input == "EXIT":
print("Exiting...", flush=True)
break
else:
# Print command
logger.typewriter_log(
"NEXT ACTION: ",
Fore.CYAN,
f"COMMAND = {Fore.CYAN}{command_name}{Style.RESET_ALL} ARGUMENTS = {Fore.CYAN}{arguments}{Style.RESET_ALL}")
# Execute command
if command_name is not None and command_name.lower().startswith("error"):
result = f"Command {command_name} threw the following error: " + arguments
elif command_name == "human_feedback":
result = f"Human feedback: {user_input}"
else:
result = f"Command {command_name} returned: {cmd.execute_command(command_name, arguments)}"
if next_action_count > 0:
next_action_count -= 1
memory_to_add = f"Assistant Reply: {assistant_reply} " \
f"\nResult: {result} " \
f"\nHuman Feedback: {user_input} "
memory.add(memory_to_add)
# Check if there's a result from the command append it to the message
# history
if result is not None:
full_message_history.append(chat.create_chat_message("system", result))
logger.typewriter_log("SYSTEM: ", Fore.YELLOW, result)
else:
full_message_history.append(
chat.create_chat_message(
"system", "Unable to execute command"))
logger.typewriter_log("SYSTEM: ", Fore.YELLOW, "Unable to execute command")
if __name__ == "__main__":
main()

15
scripts/memory/__init__.py
View File

@@ -1,12 +1,20 @@
from memory.local import LocalCache
from 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']
try:
from memory.redismem import RedisMemory
supported_memory.append('redis')
except ImportError:
print("Redis not installed. Skipping import.")
RedisMemory = None
try:
from memory.pinecone import PineconeMemory
supported_memory.append('pinecone')
except ImportError:
print("Pinecone not installed. Skipping import.")
PineconeMemory = None
@@ -28,6 +36,8 @@ def get_memory(cfg, init=False):
" use Redis as a memory backend.")
else:
memory = RedisMemory(cfg)
elif cfg.memory_backend == "no_memory":
memory = NoMemory(cfg)
if memory is None:
memory = LocalCache(cfg)
@@ -36,9 +46,14 @@ def get_memory(cfg, init=False):
return memory
def get_supported_memory_backends():
return supported_memory
__all__ = [
"get_memory",
"LocalCache",
"RedisMemory",
"PineconeMemory",
"NoMemory"
]

9
scripts/memory/base.py
View File

@@ -1,12 +1,17 @@
"""Base class for memory providers."""
import abc
from config import AbstractSingleton
from config import AbstractSingleton, Config
import openai
cfg = Config()
def get_ada_embedding(text):
text = text.replace("\n", " ")
return openai.Embedding.create(input=[text], model="text-embedding-ada-002")["data"][0]["embedding"]
if cfg.use_azure:
return openai.Embedding.create(input=[text], engine=cfg.get_azure_deployment_id_for_model("text-embedding-ada-002"))["data"][0]["embedding"]
else:
return openai.Embedding.create(input=[text], model="text-embedding-ada-002")["data"][0]["embedding"]
class MemoryProviderSingleton(AbstractSingleton):

18
scripts/memory/local.py
View File

@@ -28,10 +28,20 @@ class LocalCache(MemoryProviderSingleton):
def __init__(self, cfg) -> None:
self.filename = f"{cfg.memory_index}.json"
if os.path.exists(self.filename):
with open(self.filename, 'rb') as f:
loaded = orjson.loads(f.read())
self.data = CacheContent(**loaded)
try:
with open(self.filename, 'w+b') as f:
file_content = f.read()
if not file_content.strip():
file_content = b'{}'
f.write(file_content)
loaded = orjson.loads(file_content)
self.data = CacheContent(**loaded)
except orjson.JSONDecodeError:
print(f"Error: The file '{self.filename}' is not in JSON format.")
self.data = CacheContent()
else:
print(f"Warning: The file '{self.filename}' does not exist. Local memory would not be saved to a file.")
self.data = CacheContent()
def add(self, text: str):
@@ -54,8 +64,8 @@ class LocalCache(MemoryProviderSingleton):
vector = vector[np.newaxis, :]
self.data.embeddings = np.concatenate(
[
vector,
self.data.embeddings,
vector,
],
axis=0,
)

66
scripts/memory/no_memory.py Normal file
View File

@@ -0,0 +1,66 @@
from typing import Optional, List, Any
from memory.base import MemoryProviderSingleton
class NoMemory(MemoryProviderSingleton):
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) -> Optional[List[Any]]:
"""
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) -> Optional[List[Any]]:
"""
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 {}

11
scripts/memory/pinecone.py
View File

@@ -2,6 +2,8 @@
import pinecone
from memory.base import MemoryProviderSingleton, get_ada_embedding
from logger import logger
from colorama import Fore, Style
class PineconeMemory(MemoryProviderSingleton):
@@ -17,6 +19,15 @@ class PineconeMemory(MemoryProviderSingleton):
# 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{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)

12
scripts/memory/redismem.py
View File

@@ -7,6 +7,8 @@ from redis.commands.search.indexDefinition import IndexDefinition, IndexType
import numpy as np
from memory.base import MemoryProviderSingleton, get_ada_embedding
from logger import logger
from colorama import Fore, Style
SCHEMA = [
@@ -44,6 +46,16 @@ class RedisMemory(MemoryProviderSingleton):
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}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:

63
scripts/prompt.py Normal file
View File

@@ -0,0 +1,63 @@
from promptgenerator import PromptGenerator
def get_prompt():
"""
This function generates a prompt string that includes various constraints, commands, resources, and performance evaluations.
Returns:
str: The generated prompt string.
"""
# Initialize the PromptGenerator object
prompt_generator = PromptGenerator()
# 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 = [
("Google Search", "google", {"input": "<search>"}),
("Browse Website", "browse_website", {"url": "<url>", "question": "<what_you_want_to_find_on_website>"}),
("Start GPT Agent", "start_agent", {"name": "<name>", "task": "<short_task_desc>", "prompt": "<prompt>"}),
("Message GPT Agent", "message_agent", {"key": "<key>", "message": "<message>"}),
("List GPT Agents", "list_agents", {}),
("Delete GPT Agent", "delete_agent", {"key": "<key>"}),
("Write to file", "write_to_file", {"file": "<file>", "text": "<text>"}),
("Read file", "read_file", {"file": "<file>"}),
("Append to file", "append_to_file", {"file": "<file>", "text": "<text>"}),
("Delete file", "delete_file", {"file": "<file>"}),
("Search Files", "search_files", {"directory": "<directory>"}),
("Evaluate Code", "evaluate_code", {"code": "<full_code_string>"}),
("Get Improved Code", "improve_code", {"suggestions": "<list_of_suggestions>", "code": "<full_code_string>"}),
("Write Tests", "write_tests", {"code": "<full_code_string>", "focus": "<list_of_focus_areas>"}),
("Execute Python File", "execute_python_file", {"file": "<file>"}),
("Execute Shell Command, non-interactive commands only", "execute_shell", { "command_line": "<command_line>"}),
("Task Complete (Shutdown)", "task_complete", {"reason": "<reason>"}),
("Generate Image", "generate_image", {"prompt": "<prompt>"}),
("Do Nothing", "do_nothing", {}),
]
# Add commands to the PromptGenerator object
for command_label, command_name, args in commands:
prompt_generator.add_command(command_label, command_name, args)
# 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.")
# 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.")
# Generate the prompt string
prompt_string = prompt_generator.generate_prompt_string()
return prompt_string

129
scripts/promptgenerator.py Normal file
View File

@@ -0,0 +1,129 @@
import json
class PromptGenerator:
"""
A class for generating custom prompt strings based on constraints, commands, resources, and performance evaluations.
"""
def __init__(self):
"""
Initialize the PromptGenerator object with empty lists of constraints, commands, resources, and performance evaluations.
"""
self.constraints = []
self.commands = []
self.resources = []
self.performance_evaluation = []
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):
"""
Add a constraint to the constraints list.
Args:
constraint (str): The constraint to be added.
"""
self.constraints.append(constraint)
def add_command(self, command_label, command_name, args=None):
"""
Add a command to the commands list with a label, name, and optional arguments.
Args:
command_label (str): The label of the command.
command_name (str): The name of the command.
args (dict, optional): A dictionary containing argument names and their values. Defaults to None.
"""
if args is None:
args = {}
command_args = {arg_key: arg_value for arg_key,
arg_value in args.items()}
command = {
"label": command_label,
"name": command_name,
"args": command_args,
}
self.commands.append(command)
def _generate_command_string(self, command):
"""
Generate a formatted string representation of a command.
Args:
command (dict): A dictionary containing command information.
Returns:
str: The formatted command string.
"""
args_string = ', '.join(
f'"{key}": "{value}"' for key, value in command['args'].items())
return f'{command["label"]}: "{command["name"]}", args: {args_string}'
def add_resource(self, resource):
"""
Add a resource to the resources list.
Args:
resource (str): The resource to be added.
"""
self.resources.append(resource)
def add_performance_evaluation(self, evaluation):
"""
Add a performance evaluation item to the performance_evaluation list.
Args:
evaluation (str): The evaluation item to be added.
"""
self.performance_evaluation.append(evaluation)
def _generate_numbered_list(self, items, item_type='list'):
"""
Generate a numbered list from given items based on the item_type.
Args:
items (list): A list of items to be numbered.
item_type (str, optional): The type of items in the list. Defaults to 'list'.
Returns:
str: The formatted numbered list.
"""
if item_type == 'command':
return "\n".join(f"{i+1}. {self._generate_command_string(item)}" for i, item in enumerate(items))
else:
return "\n".join(f"{i+1}. {item}" for i, item in enumerate(items))
def generate_prompt_string(self):
"""
Generate a prompt string based on the constraints, commands, resources, and performance evaluations.
Returns:
str: The generated prompt string.
"""
formatted_response_format = json.dumps(self.response_format, indent=4)
prompt_string = (
f"Constraints:\n{self._generate_numbered_list(self.constraints)}\n\n"
f"Commands:\n{self._generate_numbered_list(self.commands, item_type='command')}\n\n"
f"Resources:\n{self._generate_numbered_list(self.resources)}\n\n"
f"Performance Evaluation:\n{self._generate_numbered_list(self.performance_evaluation)}\n\n"
f"You should only respond in JSON format as described below \nResponse Format: \n{formatted_response_format} \nEnsure the response can be parsed by Python json.loads"
)
return prompt_string

31
scripts/speak.py
View File

@@ -7,9 +7,21 @@ import gtts
import threading
from threading import Lock, Semaphore
# Default voice IDs
default_voices = ["ErXwobaYiN019PkySvjV", "EXAVITQu4vr4xnSDxMaL"]
# TODO: Nicer names for these ids
voices = ["ErXwobaYiN019PkySvjV", "EXAVITQu4vr4xnSDxMaL"]
# Retrieve custom voice IDs from the Config class
custom_voice_1 = cfg.elevenlabs_voice_1_id
custom_voice_2 = cfg.elevenlabs_voice_2_id
# Placeholder values that should be treated as empty
placeholders = {"your-voice-id"}
# Use custom voice IDs if provided and not placeholders, otherwise use default voice IDs
voices = [
custom_voice_1 if custom_voice_1 and custom_voice_1 not in placeholders else default_voices[0],
custom_voice_2 if custom_voice_2 and custom_voice_2 not in placeholders else default_voices[1]
]
tts_headers = {
"Content-Type": "application/json",
@@ -19,6 +31,7 @@ tts_headers = {
mutex_lock = Lock() # Ensure only one sound is played at a time
queue_semaphore = Semaphore(1) # The amount of sounds to queue before blocking the main thread
def eleven_labs_speech(text, voice_index=0):
"""Speak text using elevenlabs.io's API"""
tts_url = "https://api.elevenlabs.io/v1/text-to-speech/{voice_id}".format(
@@ -63,8 +76,16 @@ def gtts_speech(text):
playsound("speech.mp3", True)
os.remove("speech.mp3")
def macos_tts_speech(text):
os.system(f'say "{text}"')
def macos_tts_speech(text, voice_index=0):
if voice_index == 0:
os.system(f'say "{text}"')
else:
if voice_index == 1:
os.system(f'say -v "Ava (Premium)" "{text}"')
else:
os.system(f'say -v Samantha "{text}"')
def say_text(text, voice_index=0):
@@ -82,7 +103,7 @@ def say_text(text, voice_index=0):
success = eleven_labs_speech(text, voice_index)
if not success:
gtts_speech(text)
queue_semaphore.release()
queue_semaphore.acquire(True)

2
scripts/spinner.py
View File

@@ -20,7 +20,7 @@ class Spinner:
sys.stdout.write(next(self.spinner) + " " + self.message + "\r")
sys.stdout.flush()
time.sleep(self.delay)
sys.stdout.write('\b' * (len(self.message) + 2))
sys.stdout.write('\r' + ' ' * (len(self.message) + 2) + '\r')
def __enter__(self):
"""Start the spinner"""

4
scripts/token_counter.py
View File

@@ -1,6 +1,7 @@
import tiktoken
from typing import List, Dict
def count_message_tokens(messages : List[Dict[str, str]], model : str = "gpt-3.5-turbo-0301") -> int:
"""
Returns the number of tokens used by a list of messages.
@@ -15,7 +16,7 @@ def count_message_tokens(messages : List[Dict[str, str]], model : str = "gpt-3.5
try:
encoding = tiktoken.encoding_for_model(model)
except KeyError:
print("Warning: model not found. Using cl100k_base encoding.")
logger.warn("Warning: model not found. Using cl100k_base encoding.")
encoding = tiktoken.get_encoding("cl100k_base")
if model == "gpt-3.5-turbo":
# !Node: gpt-3.5-turbo may change over time. Returning num tokens assuming gpt-3.5-turbo-0301.")
@@ -41,6 +42,7 @@ def count_message_tokens(messages : List[Dict[str, str]], model : str = "gpt-3.5
num_tokens += 3 # every reply is primed with <|start|>assistant<|message|>
return num_tokens
def count_string_tokens(string: str, model_name: str) -> int:
"""
Returns the number of tokens in a text string.