refactor(models): Refine MessageAgentThought SQLAlchemy typing

[tomerqodo]

Benchmark PR from qodo-benchmark#425

[greptile-apps]

Greptile Summary

This PR refactors the MessageAgentThought SQLAlchemy model to use proper type annotations with Mapped[] syntax and migrates from Base to TypeBase for dataclass support.

Major Changes:

• Migrated MessageAgentThought model from Base to TypeBase with full Mapped[] type annotations following Python 3.12+ syntax
• Updated base_agent_runner.py to use Decimal types for pricing fields and CreatorUserRole enum for role fields
• Removed manual id field assignments in tests to align with SQLAlchemy's auto-generated dataclass __init__
• Added AgentThoughtValidation Pydantic model (though uses v1 syntax)
• Improved null-safety in tool input/observation parsing

Issues Found:

• Critical logic error on line 502 where tool_inputs is used instead of tool_responses for ToolPromptMessage content
• Pydantic v1 Config class used instead of v2 ConfigDict (violates compliance rule add service layer OTel Span  #14)
• Inconsistent Decimal initialization for pricing fields that may not match model defaults

Confidence Score: 2/5

• This PR has critical logic errors that will cause incorrect behavior in production
• Score reflects a critical logic bug on line 502 where tool responses are incorrectly populated with tool inputs, which will break agent tool execution flow. Additionally, inconsistent Decimal initialization could cause pricing calculation issues.
• Pay close attention to api/core/agent/base_agent_runner.py - contains logic errors in tool response handling (line 502) and pricing field initialization (lines 319-326)

Important Files Changed

Filename	Overview
api/models/model.py	Refactored MessageAgentThought from Base to TypeBase with proper type annotations using Mapped[] syntax, added proper defaults and changed base class for dataclass support
api/core/agent/base_agent_runner.py	Updated to use proper Decimal types and CreatorUserRole enum, added validation model, improved null handling, but contains logic errors in tool response handling and pricing field initialization
api/tests/test_containers_integration_tests/services/test_agent_service.py	Removed manual id assignments from test MessageAgentThought instantiations to align with SQLAlchemy dataclass auto-generation

Sequence Diagram

sequenceDiagram
    participant Runner as BaseAgentRunner
    participant DB as Database
    participant Model as MessageAgentThought
    
    Note over Runner,Model: Agent Thought Creation Flow
    Runner->>Model: create_agent_thought()
    activate Model
    Model->>Model: Initialize with Decimal types
    Model->>Model: Set created_by_role (CreatorUserRole.ACCOUNT)
    Model->>DB: Add to session
    DB-->>Runner: Return agent_thought_id
    deactivate Model
    
    Note over Runner,Model: Agent Thought Update Flow
    Runner->>DB: Query MessageAgentThought by id
    DB-->>Runner: Return agent_thought
    Runner->>Model: Update thought field (append)
    Runner->>Model: Update tool_input/observation
    Runner->>DB: Commit changes
    
    Note over Runner,Model: History Organization Flow
    Runner->>DB: Query agent_thoughts for message
    DB-->>Runner: Return list[MessageAgentThought]
    loop For each agent_thought
        Runner->>Runner: Parse tool_input JSON
        Runner->>Runner: Parse observation JSON
        Runner->>Runner: Build ToolPromptMessage list
    end
    Runner-->>Runner: Return organized prompt messages

Loading

[greptile-apps]

_{3 files reviewed, 3 comments}

_{Edit Code Review Agent Settings | Greptile}

[greptile-apps]

logic: using tool_inputs instead of tool_responses - should be tool_responses.get(tool, agent_thought.observation)

Suggested change

	content=str(tool_inputs.get(tool, agent_thought.observation)),
	content=str(tool_responses.get(tool, agent_thought.observation)),

Prompt To Fix With AI

This is a comment left during a code review.
Path: api/core/agent/base_agent_runner.py
Line: 502:502

Comment:
**logic:** using `tool_inputs` instead of `tool_responses` - should be `tool_responses.get(tool, agent_thought.observation)`

```suggestion
                                    content=str(tool_responses.get(tool, agent_thought.observation)),
```

How can I resolve this? If you propose a fix, please make it concise.

[greptile-apps]

syntax: use Pydantic v2 ConfigDict instead of nested Config class per compliance rule #14

Suggested change

	class Config:
	extra = "allow" # Pydantic v1 syntax - should use ConfigDict(extra='forbid')
	from pydantic import ConfigDict
	model_config = ConfigDict(extra='forbid')

Context Used: Context from dashboard - CLAUDE.md (source)

_{Note: If this suggestion doesn't match your team's coding style, reply to this and let me know. I'll remember it for next time!}

Prompt To Fix With AI

This is a comment left during a code review.
Path: api/core/agent/base_agent_runner.py
Line: 64:65

Comment:
**syntax:** use Pydantic v2 `ConfigDict` instead of nested `Config` class per compliance rule #14

```suggestion
    from pydantic import ConfigDict

    model_config = ConfigDict(extra='forbid')
```

**Context Used:** Context from `dashboard` - CLAUDE.md ([source](https://app.greptile.com/review/custom-context?memory=a5f96311-5bd1-49f2-9828-2ee0c089c012))

<sub>Note: If this suggestion doesn't match your team's coding style, reply to this and let me know. I'll remember it for next time!</sub>

How can I resolve this? If you propose a fix, please make it concise.

[greptile-apps]

logic: inconsistent Decimal values for pricing fields - comparing to the model definition in api/models/model.py:1865-1874, unit_price should be nullable and price_unit should have default of 0.001, but line 319-320 has unit_price=0 and price_unit=0.001, while 325-326 has them swapped

Prompt To Fix With AI

This is a comment left during a code review.
Path: api/core/agent/base_agent_runner.py
Line: 319:326

Comment:
**logic:** inconsistent Decimal values for pricing fields - comparing to the model definition in `api/models/model.py:1865-1874`, unit_price should be nullable and price_unit should have default of 0.001, but line 319-320 has unit_price=0 and price_unit=0.001, while 325-326 has them swapped

How can I resolve this? If you propose a fix, please make it concise.