# Behavioral Impact

> Removed the now-dead node-based pydantic-ai orchestrator/tool-dispatcher stack after switching the request loop to consume tinyagent events directly.

- Skill: `tools-only/behavioral-impact-3` (Agent Skill, multi-file: 3 files)
- Install (CLI): `npx skillmds@latest add tools-only/behavioral-impact-3`
- Raw SKILL.md: https://api.skillmd.com/api/skills/tools-only/behavioral-impact-3/raw
- Safety review: pending (external: skill-scanner PASS, skillspector PASS)
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: tools-only (https://skillmd.com/u/tools-only)
- Updated: 2026-09-29
- Page: https://skillmd.com/skills/tools-only/behavioral-impact-3

---


## Summary

Removed the now-dead node-based pydantic-ai orchestrator/tool-dispatcher stack after switching the request loop to consume tinyagent events directly. This reduces complexity and prevents accidental re-introduction of the legacy execution path.

## Changes

- Deleted `src/tunacode/core/agents/agent_components/orchestrator/` (tool dispatcher, usage tracker, node processor).
- Simplified exports:
  - `src/tunacode/core/agents/agent_components/__init__.py`
  - `src/tunacode/core/agents/__init__.py`
- Removed legacy tests that only validated the deleted orchestrator behavior:
  - `tests/integration/tools/test_tool_dispatcher_coverage.py`
  - `tests/integration/core/test_tool_call_lifecycle.py`
  - `tests/unit/core/test_usage_tracker.py`
- Refactored `RequestOrchestrator._run_stream()` into small event-specific handlers to satisfy Ruff complexity gates.

## Behavioral Impact

- Tool execution and streaming are now exclusively driven by tinyagent event handling.
- No backward-compatibility is provided for orchestrator-specific internals/tests.

## Related Cards

- [[tinyagent-migration]]