The current block web requests utility has a logic to avoid the system
firing into blocklisted IPs.
However, the current logic is still prone to a few security issues:
* DNS rebinding attack: due to the lack of guarantee on the used IP not
being changed during the IP checking and firing step.
* Open redirect: due to the request sensitive request headers are still
being propagated throughout the web redirect.
### Changes 🏗️
* Uses IP pinning to request the web.
* Strip `Authorization`, `Proxy-Authorization`, `Cookie` upon web
redirects.
### Checklist 📋
#### For code changes:
- [x] I have clearly listed my changes in the PR description
- [x] I have made a test plan
- [x] I have tested my changes according to the test plan:
- [x] Test the web request block, add more tests with different
validation scenarios.
(cherry picked from commit f0df4c9174)
- Related to #8784
### Changes 🏗️
- feat(frontend/library): Improve agent output styling & fix content
overflow issue
- fix(frontend/library): Fix overlap between content and inset button of
expandable input fields (#9650)
- fix(backend): Unbreak loading graph executions with missing inputs
### Checklist 📋
#### For code changes:
- [x] I have clearly listed my changes in the PR description
- [x] I have made a test plan
- [x] I have tested my changes according to the test plan:
- Run an agent with at least one input *not* filled out; view this run
in the Library
- [x] -> page should load normally
- [x] -> agent inputs should load and show normally
- Run an agent that generates long output; view this run in the Library
- [x] -> output should not overflow its container or stretch the page
layout
- [x] -> visually check that the output section looks slick
### Issue
The SendWebRequestBlock currently fails to properly route HTTP error
responses (4xx, 5xx) to their designated output pins (`client_error` and
`server_error`). Instead, these errors are being sent to the default
"Error" pin, breaking expected workflows that depend on proper error
handling.
### Root Cause
The underlying issue is that our custom `requests` module from
`backend.util.request` appears to automatically raise exceptions for
error status codes (similar to how `raise_for_status()` works in the
standard requests library). When these exceptions are thrown, the
block's conditional logic for handling different status codes is
bypassed entirely.
### Changes
This PR adds proper exception handling to catch HTTP errors raised by
the requests module and routes them to the appropriate output pins:
- Added a try-except block to capture `requests.exceptions.HTTPError`
- Extract status code and response data from the caught exception
- Yield to the proper pin based on the status code (4xx → client_error,
5xx → server_error)
- Maintain consistent behavior with the original design intent
### Additional Context
This change maintains backward compatibility while ensuring the block
behaves according to its documented functionality. Users can now
properly handle 4xx and 5xx errors in their workflows as originally
intended.
<!-- Clearly explain the need for these changes: -->
### Checklist 📋
#### For code changes:
- [x] I have clearly listed my changes in the PR description
- [x] I have made a test plan
- [x] I have tested my changes according to the test plan:
<!-- Put your test plan here: -->
- [x] Test the block with new changes and old and ensure expected
behavior
---------
Co-authored-by: Reinier van der Leer <pwuts@agpt.co>
<!-- Clearly explain the need for these changes: -->
We accidently send several emails within 10 mins and we're gonna get
blocked for spam if we keep it up
### Changes 🏗️
- moves 1 min to 60 min timer
<!-- Concisely describe all of the changes made in this pull request:
-->
### Checklist 📋
#### For code changes:
- [x] I have clearly listed my changes in the PR description
- [x] I have made a test plan
- [x] I have tested my changes according to the test plan:
<!-- Put your test plan here: -->
- [x] Test via batching a series of notiications over an hour
### Changes 🏗️
<img width="757" alt="image"
src="https://github.com/user-attachments/assets/909aab58-24c7-42ec-9580-ac3e9f32057e"
/>
Since a self-loop is now allowed for AddToListBlock, providing an entry
pin using a static output will cause infinite execution.
This PR change avoid such scenario to be allowed.
### Checklist 📋
#### For code changes:
- [x] I have clearly listed my changes in the PR description
- [x] I have made a test plan
- [x] I have tested my changes according to the test plan:
<!-- Put your test plan here: -->
- [x] Described above
Agents using Agent blocks should be seamlessly downloaded from the
marketplace to a file and imported from a file.
Requirements:
* A recursive export process that exports all the required agents to a
single file, no matter how many layers deep (taking care of potential
loops).
* An import process that expects and extracts several agents from a
single file into your library at once.
Considerations:
We need to ensure the reference IDs in the Agent Blocks match/are
updated to match the imported sub-agent ids to prevent broken
references.
### Changes 🏗️
* Add sub_graphs field on Graph model
* Improve graph creation query to support inserting graph + subgraphs in
batch
* Deprecate graph template & remove its column
* Update on marketplace download agent (unified the used method, with
more secure cleanup & proper ownership check).
* Fix failing test cases
### Checklist 📋
#### For code changes:
- [x] I have clearly listed my changes in the PR description
- [x] I have made a test plan
- [x] I have tested my changes according to the test plan:
<!-- Put your test plan here: -->
- [x] Export graph with sub agents.
- [x] Import the exported graph with sub agents.
- Resolves#9609
### Changes 🏗️
- feat(frontend/library): Add "Export agent to file" button
- fix(frontend/library): Put "Open in builder" button behind access
check
- feat(backend): Improve & move graph export stripping logic
- Add logic to strip `SecretField` values
- Move node stripping logic to `NodeModel` from `GraphModel`
- Add `NodeModel.stripped_for_export()` method
- Add `NodeModel.block` property
### Checklist 📋
#### For code changes:
- [x] I have clearly listed my changes in the PR description
- [x] I have made a test plan
- [x] I have tested my changes according to the test plan:
- Create and configure an agent with the Publish To Medium block and a
block that uses credentials
- Go to `/library/agents/[id]` for the agent you just created
- [x] -> "Open in builder" button should show
- [x] -> "Open in builder" button should work
- [x] -> "Export agent to file" button should show
- [x] -> "Export agent to file" button should work
- [x] -> Exported file contains no credentials or secrets
- [ ] -> ~~Exported file contains no user IDs~~
- Go to `/library/agents/[id]` for an agent from the marketplace
- [x] -> "Open in builder" button should not show
- [x] -> "Export agent to file" button should not show
Although returning a Prisma object on an RPC is a bad practice, we have
instances where we do so and the type contains a `prisma.Json` field.
This Json field can't be seamlessly serialized and then converted back
into the Prisma object.
### Changes 🏗️
Replacing prisma object as return type on notification service with a
plain pydantic object as DTO.
### Checklist 📋
#### For code changes:
- [x] I have clearly listed my changes in the PR description
- [x] I have made a test plan
- [x] I have tested my changes according to the test plan:
- [x] Calling notification APIs through the RPC client.
(cherry picked from commit b9f31a9c44)
DatabaseManager is already provisioned in RestApiService, and
NotificationService lives within the same instance as the Rest Server.
### Changes 🏗️
Moving the DB calls of NotificationService to DatabaseManager.
### Checklist 📋
#### For code changes:
- [ ] I have clearly listed my changes in the PR description
- [ ] I have made a test plan
- [ ] I have tested my changes according to the test plan:
<!-- Put your test plan here: -->
- [ ] ...
<details>
<summary>Example test plan</summary>
- [ ] Create from scratch and execute an agent with at least 3 blocks
- [ ] Import an agent from file upload, and confirm it executes
correctly
- [ ] Upload agent to marketplace
- [ ] Import an agent from marketplace and confirm it executes correctly
- [ ] Edit an agent from monitor, and confirm it executes correctly
</details>
#### For configuration changes:
- [ ] `.env.example` is updated or already compatible with my changes
- [ ] `docker-compose.yml` is updated or already compatible with my
changes
- [ ] I have included a list of my configuration changes in the PR
description (under **Changes**)
<details>
<summary>Examples of configuration changes</summary>
- Changing ports
- Adding new services that need to communicate with each other
- Secrets or environment variable changes
- New or infrastructure changes such as databases
</details>
(cherry picked from commit f4d4bb83b0)
### Changes 🏗️
This is a follow-up of
https://github.com/Significant-Gravitas/AutoGPT/pull/9610
* Addressing the PR comments described in the mentioned PR
* Removed debug logging
* Fix image state loading logic on agent upload process
### Checklist 📋
#### For code changes:
- [x] I have clearly listed my changes in the PR description
- [x] I have made a test plan
- [x] I have tested my changes according to the test plan:
- [x] Create an library agent and try to create a store listing.
<img width="1409" alt="image"
src="https://github.com/user-attachments/assets/dc86dc97-a33f-4336-ab90-19a53c6f7e0f"
/>
<!-- Clearly explain the need for these changes: -->
We want to be able to process emails on a scheduled basis for summaries.
This adds the baselines for that
### Changes 🏗️
- Adds new tooling to Scheduluer to handle the in-memory schedule for
the weekly processing
- Adds new exposes to notification manager to handle the different data
models for scheduled emails
- adds new models to the notification data models to handle the
different requirements for scheduled emails, closely paralleling the
existing notification ones
- Adds new email template
Note: After testing, email sending was disabled until the template and
data filling are done later down the line. We don't want to email people
random stuff, ya know?
<!-- Concisely describe all of the changes made in this pull request:
-->
### Checklist 📋
#### For code changes:
- [x] I have clearly listed my changes in the PR description
- [x] I have made a test plan
- [x] I have tested my changes according to the test plan:
<!-- Put your test plan here: -->
- [x] Test sending an email on the scheduled basis
- [x] Make sure you get the email, ignoring the fact that all the data
isn't real inside it
---------
Co-authored-by: Zamil Majdy <zamil.majdy@agpt.co>
### Changes 🏗️
We've been auto-generating the thumbnail image for the agent in the
library, this PR is propagating that image as an initial image for the
store listing. This PR also removes the fetch all query for getting the
count for paginating library agent.
### Checklist 📋
#### For code changes:
- [x] I have clearly listed my changes in the PR description
- [x] I have made a test plan
- [x] I have tested my changes according to the test plan:
- [x] Create an library agent and try to create a store listing.
<img width="1409" alt="image"
src="https://github.com/user-attachments/assets/dc86dc97-a33f-4336-ab90-19a53c6f7e0f"
/>
### Changes 🏗️
Removed some indices in the marketplace table that were useless.
### Checklist 📋
#### For code changes:
- [ ] I have clearly listed my changes in the PR description
- [ ] I have made a test plan
- [ ] I have tested my changes according to the test plan:
<!-- Put your test plan here: -->
- [ ] ...
<details>
<summary>Example test plan</summary>
- [ ] Create from scratch and execute an agent with at least 3 blocks
- [ ] Import an agent from file upload, and confirm it executes
correctly
- [ ] Upload agent to marketplace
- [ ] Import an agent from marketplace and confirm it executes correctly
- [ ] Edit an agent from monitor, and confirm it executes correctly
</details>
#### For configuration changes:
- [ ] `.env.example` is updated or already compatible with my changes
- [ ] `docker-compose.yml` is updated or already compatible with my
changes
- [ ] I have included a list of my configuration changes in the PR
description (under **Changes**)
<details>
<summary>Examples of configuration changes</summary>
- Changing ports
- Adding new services that need to communicate with each other
- Secrets or environment variable changes
- New or infrastructure changes such as databases
</details>
<!-- Clearly explain the need for these changes: -->
We don't want to spam the user with similar notification types so we
want to group them up over a timespan and handle that as a group of
notifications.
### Changes 🏗️
- Adds a batch queue
- Moves the ExecutionScheduleur to a generic Scheduler
- Makes the Agent run a batch operation
- Fixes various bugs in how we originally made the batch db models and
queries
<!-- Concisely describe all of the changes made in this pull request:
-->
### Checklist 📋
#### For code changes:
- [x] I have clearly listed my changes in the PR description
- [x] I have made a test plan
- [x] I have tested my changes according to the test plan:
<!-- Put your test plan here: -->
- [x] Run 10 agents back to back
- [x] Notice how many emails you get
- [x] Wait a bit and you should after an hour (change the cron rule to
speed up testing this) you'll get an email and see all the batches in
your db are empty
---------
Co-authored-by: Zamil Majdy <zamil.majdy@agpt.co>
<!-- Clearly explain the need for these changes: -->
When swapping branches back and forth, I often hit stuff in my db that
is incompatible with other branches. This protects against it a bit
more.
### Changes 🏗️
- Adds filtering around converting data out of the db
<!-- Concisely describe all of the changes made in this pull request:
-->
### Checklist 📋
#### For code changes:
- [x] I have clearly listed my changes in the PR description
- [x] I have made a test plan
- [x] I have tested my changes according to the test plan:
<!-- Put your test plan here: -->
- [x] Test with weird and broken agents in db
- [x] Go to the library page
- [x] Notice in the server console that it warns of bad agents and the
page still loads
It's hard to configure the DB parameter at the moment, this PR gives an
option to provide URL parameter on the Prisma DATABASE_URL.
### Changes 🏗️
Changes:
* Reduce excessive application log
* Make DB configurable
* Set DB connection limit & timeout
### Checklist 📋
#### For code changes:
- [x] I have clearly listed my changes in the PR description
- [x] I have made a test plan
- [x] I have tested my changes according to the test plan:
### Changes 🏗️
Creating a self-loop of the AddToListBlock requires a boilerplate of
adding a dummy block to trigger its execution.
If the only inbound link on an input pin comes from the same block, we
don't wait for that pin on the first execution (as it's never going to
come).
### Checklist 📋
#### For code changes:
- [ ] I have clearly listed my changes in the PR description
- [ ] I have made a test plan
- [ ] I have tested my changes according to the test plan:
<!-- Put your test plan here: -->
- [ ] ...
<details>
<summary>Example test plan</summary>
- [ ] Create from scratch and execute an agent with at least 3 blocks
- [ ] Import an agent from file upload, and confirm it executes
correctly
- [ ] Upload agent to marketplace
- [ ] Import an agent from marketplace and confirm it executes correctly
- [ ] Edit an agent from monitor, and confirm it executes correctly
</details>
#### For configuration changes:
- [ ] `.env.example` is updated or already compatible with my changes
- [ ] `docker-compose.yml` is updated or already compatible with my
changes
- [ ] I have included a list of my configuration changes in the PR
description (under **Changes**)
<details>
<summary>Examples of configuration changes</summary>
- Changing ports
- Adding new services that need to communicate with each other
- Secrets or environment variable changes
- New or infrastructure changes such as databases
</details>
addresses the changes requested here [Move to a single source of truth
for docs — Remove duplicate info from readme
#8887](https://github.com/Significant-Gravitas/AutoGPT/issues/8887)
### Changes 🏗️
<!-- Concisely describe all of the changes made in this pull request:
-->
changed readme files
<!-- Clearly explain the need for these changes: -->
### Changes 🏗️
The Exa Block has a missing error pin which prevents error handling,
this adds that.
<!-- Clearly explain the need for these changes: -->
Per chat with toran, we want one click unsubscribe to work in things
like gmail to reduce spam. This implements the one click unsubscribe
feature, but it is difficult to test due to not being able to control
when it is shown.
### Changes 🏗️
- Adds one click unsub
- Casually fix google reauth by checking the correct variables (approved
to be in pr by Toran)
<!-- Concisely describe all of the changes made in this pull request:
-->
### Checklist 📋
#### For code changes:
- [x] I have clearly listed my changes in the PR description
- [x] I have made a test plan
- [x] I have tested my changes according to the test plan:
<!-- Put your test plan here: -->
- [x] Send an email
- [x] Open headers on email and pull the link out
- [x] `curl -X POST <link>` and ensure unsubscribed from all messages.
Note that we can't control when the box is shown on various providers so
we just verify it works on our side for now by checking the headers
#### Configuration changes
Infra pr coming separately, but we did add the secret default to the
.env.example and the docker compose
---------
Co-authored-by: Zamil Majdy <zamil.majdy@agpt.co>
### Changes 🏗️
This is to add the Otto chat bot to the frontend UI.
The changes i have made for this goes as follows:
- I have made and added a basic chat UI widget to the build page, this
will become the main UI for Otto, this will be getting updates in the
future.
- I have made an API route ``/api/otto/ask`` that will receive the
message from the frontend and pass it to the Otto API that we have setup
on GCP.
I already have plans for how we can improve the UI and make it more
modular, this way we can add more features to Otto down the line!
### Config Changes
When we merge this, we will need to add the ENV var for ``otto_api_url``
that will be the URL from GCP where the API is being hosted!
This is now ready for a review as the backend API is up and running, to
who ever does test this, if you want the API url please DM me 😄
### Checklist 📋
#### For code changes:
- [x] I have clearly listed my changes in the PR description
- [x] I have made a test plan
- [x] I have tested my changes according to the test plan:
<!-- Put your test plan here: -->
- [x] Set the ENV var for ``otto_api_url`` - ask @Bentlybro for this
- [x] Set the ENV var for ``NEXT_PUBLIC_BEHAVE_AS`` to ``CLOUD``
- [x] Send otto a message from the build page frontend
- [x] Send otto a message that contains graph data to see if the sending
graph data works
This PR adds backend to make Onboarding UI added in
https://github.com/Significant-Gravitas/AutoGPT/pull/9485 functional and
adds missing confetti screen at the end of Onboarding.
*Make sure to have at least any 2 agents in store when testing locally.*
Otherwise the onboarding will finish without showing agents.
Visit `/onboarding/reset` to reset onboarding state, otherwise it'll
always redirect to `/library` once finished.
### Changes 🏗️
- Onboarding opens automatically on sign up and login (if unfinished)
for all users
- Update db schema to add `UserOnboarding` and add migration
- Add GET and PATCH `/onboarding` endpoints and logic to retrieve and
update data Onboarding for a user
- Update `POST /library/agents` endpoint
(`addMarketplaceAgentToLibrary`), so it includes input and output nodes;
these are needed to know input schema for the Onboarding
- Use new endpoints during onboarding to fetch and update data
- Use agents from the marketplace and their input schema for the
onboarding
- Add algorithm to choose store agents based on user answers and related
endpoint `GET /onboarding/agents`
- Redirect outside onboarding if finished and resume on proper page
- Add `canvas-confetti` and `@types/canvas-confetti` frontend packages
- Add and show congrats confetti screen when user runs and agent and
before opening library
- Minor design updates and onboarding fixes
### Checklist 📋
#### For code changes:
- [x] I have clearly listed my changes in the PR description
- [x] I have made a test plan
- [x] I have tested my changes according to the test plan:
- [x] Redirect to `/onboarding/*` on sign up and sign in (if onboarding
unfinished)
- [x] Onboarding works and can be finished
- [x] Agent runs on finish
- [x] Onboarding state is saved and logging out or refreshing page
restores correct state (user choices)
- [x] When onboarding finished, trying to go into `/onboarding`
redirects to `/library`
---------
Co-authored-by: Nicholas Tindle <nicholas.tindle@agpt.co>
Co-authored-by: Reinier van der Leer <pwuts@agpt.co>
### Changes 🏗️
There are a few agent-loop issues that this PR is addressing:
* There is a lack of support for agent-loop in Anthropic.
* Duplicated system & user prompt as the main objective prompt in the
agent loop.
* A long rendered text of conversation history by
SmartDecisionMakerBlock agent-loop in the UI.
* A lack of execution input being rendered in the execution list making
it harder to debug.
https://github.com/user-attachments/assets/be430000-bde0-40c6-8f2e-c97ce45b5ed1
### Checklist 📋
#### For code changes:
- [x] I have clearly listed my changes in the PR description
- [x] I have made a test plan
- [x] I have tested my changes according to the test plan:
<!-- Put your test plan here: -->
- [x] Create from scratch and execute an agent with at least 3 blocks
using SmartDecisionMaker Block.
- Fixes#9577
### Changes 🏗️
- Add the required include statements to `LibraryAgent` queries
- Make a `library_agent_include(user_id)` utility in
`backend.data.includes`
### Checklist 📋
#### For code changes:
- [x] I have clearly listed my changes in the PR description
- [x] I have made a test plan
- [ ] I have tested my changes according to the test plan:
- Go to an agent in the Marketplace; click "Add To Library"
- [ ] -> should work, not return an error
- [ ] -> should redirect to `/library/agents/[id]` -> should load
normally
<!-- Clearly explain the need for these changes: -->
As part of the code review agent, I found these blocks would be useful,
but not required. Adds the ability to list comments on an issue and PR
as well as update comments on an issue, pr and discussion
### Changes 🏗️
<!-- Concisely describe all of the changes made in this pull request:
-->
- Adds Listing comments block
- Adds updating comment block
### Checklist 📋
#### For code changes:
- [x] I have clearly listed my changes in the PR description
- [x] I have made a test plan
- [x] I have tested my changes according to the test plan:
<!-- Put your test plan here: -->
- [x] Tested all of the new blocks on #9582 . Review it for context
---------
Co-authored-by: Copilot Autofix powered by AI <62310815+github-advanced-security[bot]@users.noreply.github.com>
Co-authored-by: Bentlybro <tomnoon9@gmail.com>
<!-- Clearly explain the need for these changes: -->
Trying to run the GitHub Status blocks results in a serialization error
### Changes 🏗️
- Serializes the checks and blocks data objects correctly
<!-- Concisely describe all of the changes made in this pull request:
-->
### Checklist 📋
#### For code changes:
- [x] I have clearly listed my changes in the PR description
- [x] I have made a test plan
- [x] I have tested my changes according to the test plan:
<!-- Put your test plan here: -->
- [x] make a one block agent, test it and make sure it works
- Resolves#9542
### Changes 🏗️
- feat(platform): Add "Delete run" button on `/library/agents/[id]` w/
confirm dialog
- Add `AgentGraphExecution.isDeleted` column for soft delete
- Add `DELETE /api/executions/{graph_exec_id}` endpoint
- feat(frontend): Add "Stop run" button on `/library/agents/[id]`
Technical improvements:
- refactor(frontend): Generalize `AgentDeleteConfirmDialog` ->
`DeleteConfirmDialog`
- refactor(frontend): Brand `GraphExecution.execution_id` and
`Schedule.id` to prevent mixing
### Checklist 📋
#### For code changes:
- [x] I have clearly listed my changes in the PR description
- [x] I have made a test plan
- [x] I have tested my changes according to the test plan:
- Click "Delete run" under "Run actions" in the right sidebar
- [x] -> Confirmation dialog should show
- Click "Delete"
- [x] -> run should disappear from list
- [x] -> view should switch to "Draft new run" view
- [x] -> refresh page -> run should not be in list
- Click "Delete" in the menu on a run in the list
- [x] -> Confirmation dialog should show
- Click "Delete"
- [x] -> run should disappear from list
- [x] -> refresh page -> run should not be in list
---------
Co-authored-by: Krzysztof Czerwinski <34861343+kcze@users.noreply.github.com>
### Changes 🏗️
Integrate Ideogram for auto-generating created agent thumbnail
**Note:** switching back to the "old" Replicate-based image generator is
possible by setting `USE_AGENT_IMAGE_GENERATION_V2=false`.
### Checklist 📋
#### For code changes:
- [ ] I have clearly listed my changes in the PR description
- [ ] I have made a test plan
- [ ] I have tested my changes according to the test plan:
<!-- Put your test plan here: -->
- [ ] ...
<details>
<summary>Example test plan</summary>
- [ ] Create from scratch and execute an agent with at least 3 blocks
- [ ] Import an agent from file upload, and confirm it executes
correctly
- [ ] Upload agent to marketplace
- [ ] Import an agent from marketplace and confirm it executes correctly
- [ ] Edit an agent from monitor, and confirm it executes correctly
</details>
---------
Co-authored-by: Reinier van der Leer <pwuts@agpt.co>
Enhance process management with error handling, lifecycle improvements,
and better resource management
- Added robust error handling for process startup failures.
- Implemented proper cleanup of processes during failure scenarios.
- Added detailed error logging for process failures and cleanup
operations.
- Enhanced process lifecycle with health checks.
- Added success logging for process terminations.
These updates enhance the reliability and maintainability of the system.
---------
Co-authored-by: Zamil Majdy <zamil.majdy@agpt.co>
### Changes 🏗️
Prevent a graph being executed and do nothing by erroring out when no
starting node is available.
### Checklist 📋
#### For code changes:
- [ ] I have clearly listed my changes in the PR description
- [ ] I have made a test plan
- [ ] I have tested my changes according to the test plan:
<!-- Put your test plan here: -->
- [ ] ...
<details>
<summary>Example test plan</summary>
- [ ] Create from scratch and execute an agent with at least 3 blocks
- [ ] Import an agent from file upload, and confirm it executes
correctly
- [ ] Upload agent to marketplace
- [ ] Import an agent from marketplace and confirm it executes correctly
- [ ] Edit an agent from monitor, and confirm it executes correctly
</details>
#### For configuration changes:
- [ ] `.env.example` is updated or already compatible with my changes
- [ ] `docker-compose.yml` is updated or already compatible with my
changes
- [ ] I have included a list of my configuration changes in the PR
description (under **Changes**)
<details>
<summary>Examples of configuration changes</summary>
- Changing ports
- Adding new services that need to communicate with each other
- Secrets or environment variable changes
- New or infrastructure changes such as databases
</details>
We've disabled the request for security reasons, and the current request
library breaks on the 301 web response.
### Changes 🏗️
Add manually safe URL redirect on our web request layer.
### Checklist 📋
#### For code changes:
- [ ] I have clearly listed my changes in the PR description
- [ ] I have made a test plan
- [ ] I have tested my changes according to the test plan:
<!-- Put your test plan here: -->
- [ ] ...
<details>
<summary>Example test plan</summary>
- [ ] Create from scratch and execute an agent with at least 3 blocks
- [ ] Import an agent from file upload, and confirm it executes
correctly
- [ ] Upload agent to marketplace
- [ ] Import an agent from marketplace and confirm it executes correctly
- [ ] Edit an agent from monitor, and confirm it executes correctly
</details>
#### For configuration changes:
- [ ] `.env.example` is updated or already compatible with my changes
- [ ] `docker-compose.yml` is updated or already compatible with my
changes
- [ ] I have included a list of my configuration changes in the PR
description (under **Changes**)
<details>
<summary>Examples of configuration changes</summary>
- Changing ports
- Adding new services that need to communicate with each other
- Secrets or environment variable changes
- New or infrastructure changes such as databases
</details>
<!-- Clearly explain the need for these changes: -->
For emailing, we want the user to know when an agent stopped because
their balance was too low. This is the first step of that.
### Changes 🏗️
- Raise InsufficientBalanceError from credit system rather than value
error when user runs out of money
- Handle when an agent output isn't hooked up well
- Fix the contents of the email for low balance to be a bit more aligned
with the PRD
- expose the topup intent from the db manager
- objectify the execution stats so we can pass it around a bit more type
safe
- extract the notification stuff in manager into a function
<!-- Concisely describe all of the changes made in this pull request:
-->
### Checklist 📋
#### For code changes:
- [x] I have clearly listed my changes in the PR description
- [x] I have made a test plan
- [x] I have tested my changes according to the test plan:
<!-- Put your test plan here: -->
- [x] Set balance to $0.01
- [x] Run an agent that costs something more than $0.01
- [x] Check you get an email
- [x] Check your top up link works
---------
Co-authored-by: Zamil Majdy <zamil.majdy@agpt.co>
- Resolves#9545
### Changes 🏗️
- fix(platform): Make "Delete" button on `/monitoring` soft-delete the
`LibraryAgent` instead of hard-deleting the corresponding `AgentGraph`
- feat(frontend): Add "Delete agent" button on `/library/agents/[id]`
Technical:
- fix(backend): Accept partial input on `update_library_agent` endpoint
- feat(backend): Add `GET /api/library/agents/{library_agent_id}`
endpoint
- refactor(frontend): Replace use of `GraphMeta` by `LibraryAgent` where
possible on `/library/agents/[id]`
Also, out of scope but important:
- fix(frontend): Hide buttons that require direct graph access depending
on `agent.can_access_graph`
### Checklist 📋
#### For code changes:
- [x] I have clearly listed my changes in the PR description
- [x] I have made a test plan
- [x] I have tested my changes according to the test plan:
- Go to the Library and select an agent
- [x] -> `/library/agents/[id]` should load normally
- Save this URL for later (especially the ID)!
- Click "Delete agent" on `/library/agents/[id]`
- [x] -> should show a confirmation dialog
- Click "Delete" to confirm
- [x] -> should redirect back to `/library`
- [x] -> deleted agent should no longer be listed in the Library
- Click "Delete agent" on `/monitoring`
- [x] -> should show a confirmation dialog
- Click "Delete" to confirm
- [x] -> agent should disappear from agent list
- [x] -> views should reset / deselect the deleted agent where
applicable
**Proposal / Request For Comments**
The tool calling provided by the LLM provider requires an output
generated by the tool to be looped back as part of the conversation
history. The scope of this PR is trying to address the need to fulfill
the feature expectation.
### Changes 🏗️
* `Last Tool Output` is introduced to loop back the output of the tool
back to Smart Decision Block.
* Smart Decision Block execution will be pending unless the `Last Tool
Output` us provided where a pending tool call is present in the
conversation history.
* **Known hack**: The last tool output will prefill all the pending tool
calls from the conversation history.
* A few tweaks were required to allow a blocking loop to be executed
without awaiting on all the inbound links.
<img width="1395" alt="image"
src="https://github.com/user-attachments/assets/fdad4407-621b-45d0-a457-76b2d4c853b9"
/>
### Checklist 📋
#### For code changes:
- [ ] I have clearly listed my changes in the PR description
- [ ] I have made a test plan
- [ ] I have tested my changes according to the test plan:
<!-- Put your test plan here: -->
- [ ] ...
<details>
<summary>Example test plan</summary>
- [ ] Create from scratch and execute an agent with at least 3 blocks
- [ ] Import an agent from file upload, and confirm it executes
correctly
- [ ] Upload agent to marketplace
- [ ] Import an agent from marketplace and confirm it executes correctly
- [ ] Edit an agent from monitor, and confirm it executes correctly
</details>
#### For configuration changes:
- [ ] `.env.example` is updated or already compatible with my changes
- [ ] `docker-compose.yml` is updated or already compatible with my
changes
- [ ] I have included a list of my configuration changes in the PR
description (under **Changes**)
<details>
<summary>Examples of configuration changes</summary>
- Changing ports
- Adding new services that need to communicate with each other
- Secrets or environment variable changes
- New or infrastructure changes such as databases
</details>
### Changes 🏗️
Instead of parsing an LlmMessage by ourselves, we use raw content from
the LLM as conversation history and accept any format it introduces.
Example output:
```
[
{
"role": "system",
"content": "Thinking carefully step by step decide which function to call. Always choose a function call from the list of function signatures. The test graph is basically an all knowing answer machine you can use it to get an answer"
},
{
"role": "user",
"content": "Hey how's the weather today"
},
{
"role": "assistant",
"audio": null,
"content": null,
"refusal": null,
"tool_calls": [
{
"id": "call_Z7CKKIkldylmfWJdE6ZnDxjr",
"type": "function",
"function": {
"name": "storevalueblock",
"arguments": "{\"input\":\"I don't have context for your location. Could you provide one?\"}"
}
}
],
"function_call": null
}
]
```
### Checklist 📋
#### For code changes:
- [ ] I have clearly listed my changes in the PR description
- [ ] I have made a test plan
- [ ] I have tested my changes according to the test plan:
<!-- Put your test plan here: -->
- [ ] ...
<details>
<summary>Example test plan</summary>
- [ ] Create from scratch and execute an agent with at least 3 blocks
- [ ] Import an agent from file upload, and confirm it executes
correctly
- [ ] Upload agent to marketplace
- [ ] Import an agent from marketplace and confirm it executes correctly
- [ ] Edit an agent from monitor, and confirm it executes correctly
</details>
#### For configuration changes:
- [ ] `.env.example` is updated or already compatible with my changes
- [ ] `docker-compose.yml` is updated or already compatible with my
changes
- [ ] I have included a list of my configuration changes in the PR
description (under **Changes**)
<details>
<summary>Examples of configuration changes</summary>
- Changing ports
- Adding new services that need to communicate with each other
- Secrets or environment variable changes
- New or infrastructure changes such as databases
</details>
