refactor(models): Refine MessageAgentThought SQLAlchemy typing#162
Conversation
Review Summary by QodoRefine MessageAgentThought SQLAlchemy typing and improve agent data handling
WalkthroughsDescription• Refine MessageAgentThought SQLAlchemy model with proper type annotations - Convert to TypeBase inheritance for consistency - Add Mapped type hints for all columns - Use Decimal type for numeric price fields • Improve agent thought data handling and validation - Add AgentThoughtValidation Pydantic model for data validation - Fix string concatenation logic in save_agent_thought method - Use CreatorUserRole enum instead of string literal • Enhance tool data processing in organize_agent_history method - Improve null-safety checks for tool input and observation payloads - Add explicit null handling before JSON parsing - Fix tool response content assignment bug • Update test fixtures to remove manual id assignment - Leverage auto-generated UUID defaults in model Diagramflowchart LR
A["MessageAgentThought Model"] -->|"Add Mapped type hints"| B["Proper SQLAlchemy Typing"]
A -->|"Use Decimal type"| C["Numeric Price Fields"]
A -->|"Inherit TypeBase"| D["Consistent Base Class"]
E["Agent Thought Processing"] -->|"Add validation model"| F["AgentThoughtValidation"]
E -->|"Improve null-safety"| G["Better Error Handling"]
E -->|"Fix string logic"| H["Correct Concatenation"]
I["Test Fixtures"] -->|"Remove manual IDs"| J["Use Auto-Generated UUIDs"]
File Changes1. api/models/model.py
|
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
+52
to
+66
| class AgentThoughtValidation(BaseModel): | ||
| """ | ||
| Validation model for agent thought data before database persistence. | ||
| """ | ||
|
|
||
| message_id: str | ||
| position: int | ||
| thought: str | None = None | ||
| tool: str | None = None | ||
| tool_input: str | None = None | ||
| observation: str | None = None | ||
|
|
||
| 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 ✓ Correctness
AgentThoughtValidation uses Pydantic v1-style class Config and allows extra fields, which weakens validation and violates the required Pydantic v2 configuration pattern. This can permit unexpected/unvalidated data to pass through persistence logic.
Agent Prompt
## Issue description
`AgentThoughtValidation` is implemented with Pydantic v1-style `class Config` and sets `extra = "allow"`, which violates the requirement to use Pydantic v2 configuration (`ConfigDict`) and to forbid undeclared fields by default.
## Issue Context
The compliance requirement mandates Pydantic v2 patterns for DTO/data validation and `ConfigDict(extra='forbid')` by default to prevent unexpected fields.
## 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
499
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.
2. Tool output uses inputs 🐞 Bug ✓ Correctness
BaseAgentRunner.organize_agent_history populates ToolPromptMessage.content from tool_inputs instead of the tool observation payload, so reconstructed history feeds the LLM the tool arguments rather than the tool results. This can break follow-up reasoning because prior tool call outputs are missing/incorrect in the prompt history.
Agent Prompt
### Issue description
`BaseAgentRunner.organize_agent_history()` builds `ToolPromptMessage` instances with `content` sourced from `tool_inputs` (the tool arguments) instead of `tool_responses` (the tool observations/outputs). This causes reconstructed conversation history to provide the LLM incorrect tool-role messages.
### Issue Context
Elsewhere, tool-role messages are constructed from tool invocation results (see `fc_agent_runner.py`), and `MessageAgentThought.observation` is treated as tool outputs (see `tool_outputs_dict` and `AgentService.get_agent_logs`).
### Fix Focus Areas
- api/core/agent/base_agent_runner.py[487-506]
### Suggested change
Replace:
- `content=str(tool_inputs.get(tool, agent_thought.observation))`
With something equivalent to:
- `content=str(tool_responses.get(tool, agent_thought.observation))`
Optionally keep the same fallback behavior if `tool_responses` is not a dict (defensive cast) and ensure `tool_responses` is always defined for the loop.
ⓘ 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