refactor(models): Refine MessageAgentThought SQLAlchemy typing#149
Conversation
Code Review by Qodo
1. Pydantic v1 Config used
|
![qodo-code-review[bot]](https://avatars.githubusercontent.com/in/484649?s=60&v=4){kind=small}
Comment on lines
+64
to
+65
| class Config: | ||
| extra = "allow" # Pydantic v1 syntax - should use ConfigDict(extra='forbid') |
There was a problem hiding this comment.
1. Pydantic v1 config used 📘 Rule violation ⛨ Security
• AgentThoughtValidation defines class Config with extra = "allow", which is Pydantic v1-style configuration and permits undeclared fields to pass validation. • This violates the requirement to use Pydantic v2 configuration patterns (ConfigDict(extra='forbid')) and weakens validation guarantees for data persisted to the database.
Agent prompt
## Issue description
`AgentThoughtValidation` uses Pydantic v1-style `class Config` and allows extra fields (`extra = "allow"`), violating the project requirement to use Pydantic v2 patterns and forbid undeclared fields by default.
## Issue Context
The compliance rule requires Pydantic v2 configuration via `ConfigDict(extra='forbid')` (and v2 validator patterns).
## Fix Focus Areas
- api/core/agent/base_agent_runner.py[52-66]
ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools
Comment on lines
+471
to
+483
| try: | ||
| tool_inputs = json.loads(tool_input_payload) | ||
| except Exception: | ||
| tool_inputs = {tool: {} for tool in tool_names} | ||
| else: | ||
| tool_inputs = {tool: {} for tool in tool_names} | ||
|
|
||
| observation_payload = agent_thought.observation | ||
| if observation_payload: | ||
| try: | ||
| tool_responses = json.loads(observation_payload) | ||
| except Exception: | ||
| tool_responses = dict.fromkeys(tool_names, observation_payload) |
There was a problem hiding this comment.
2. Swallowed json parse exceptions 📘 Rule violation ✧ Quality
• json.loads() failures for tool_input_payload and observation_payload are caught with broad except Exception and silently replaced with defaults. • This hides malformed data problems (no actionable context/logging), making production debugging and edge-case diagnosis significantly harder.
Agent prompt
## Issue description
The code broadly catches `Exception` during JSON parsing and silently falls back to defaults without logging, which swallows errors and removes critical debugging context.
## Issue Context
This runs while building agent history; malformed JSON should be handled gracefully but still produce actionable logs with identifiers.
## Fix Focus Areas
- api/core/agent/base_agent_runner.py[469-485]
ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools
Comment on lines
500
to
505
| tool_call_response.append( | ||
| ToolPromptMessage( | ||
| content=tool_responses.get(tool, agent_thought.observation), | ||
| content=str(tool_inputs.get(tool, agent_thought.observation)), | ||
| name=tool, | ||
| tool_call_id=tool_call_id, | ||
| ) |
There was a problem hiding this comment.
3. Tool output uses inputs 🐞 Bug ✓ Correctness
• ToolPromptMessage.content is populated from parsed tool *inputs* rather than parsed tool *responses/observation*, so the model will “see” arguments as if they were tool results. • This breaks tool-call chains (the assistant cannot incorporate real tool outputs) and can cause incorrect reasoning/actions across all agent conversations that rely on history replay.
Agent prompt
### Issue description
`organize_agent_history` emits TOOL-role messages with the tool **inputs** instead of the tool **outputs**. This corrupts the reconstructed conversation history and breaks model tool-call reasoning.
### Issue Context
- `tool_inputs` comes from `agent_thought.tool_input`.
- `tool_responses` comes from `agent_thought.observation`.
- `ToolPromptMessage.content` should represent the tool result and should be a string (or multimodal content).
### Fix Focus Areas
- api/core/agent/base_agent_runner.py[460-506]
- api/core/model_runtime/entities/message_entities.py[143-151]
ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Benchmark PR from agentic-review-benchmarks#11