2158 2026 02 04 Main Agent Refactoring C3de294f

Main Agent Module Refactoring

Summary

Refactored src/tunacode/core/agents/main.py from 641 lines to 389 lines (39% reduction) by extracting focused modules and removing dead code.

Changes

New Files Created

1. history_preparer.py - HistoryPreparer class

   • Encapsulates message history preparation logic
   • Handles pruning, cleanup, and sanitization
   • Single prepare() method returns ready-to-use history

2. request_logger.py - Logging utilities

   • log_history_state() - Logs message history for diagnostics
   • log_sanitized_history_state() - Logs sanitized history in debug mode
   • log_run_handle_context_messages() - Logs pydantic-ai context
   • log_node_details() - Logs node type information
   • message_has_tool_calls() - Helper for detecting tool calls

Removed Dead Code

• DotDict class - Unused utility class
• colors variable - Never referenced
• IterationManager class - Trivially simple, inlined into _handle_iteration_node()

Cleaned Up

• EmptyResponseHandler.prompt_action() - Replaced inline StateProxy class with @dataclass StateView
• Simplified counter tracking logic in EmptyResponseHandler.track()

Rationale

The original main.py had accumulated multiple responsibilities:

• Request initialization
• Message history preparation (7+ methods)
• Logging (7+ methods)
• Agent iteration loop
• Node handling
• Abort cleanup

This violated single responsibility principle and made the file difficult to navigate.

Architecture Compliance

All new modules follow layer dependency rules:

• core modules only import from tools, utils, types, and infrastructure
• No imports from ui layer

Verified by passing tests/architecture/test_layer_dependencies.py.

Updated Documentation

• docs/codebase-map/modules/core-agents.md - Updated exports and component list
• docs/architecture/agent-loop.md - Updated line references and file list