delegate-to-ailisted

# Delegate to External AI Routes tasks to specialized models based on task type using Bifrost (single-model) or PAL MCP (multi-model). ## When to Delegate Delegate when Claude is not the best tool: - **Large context** (1M+ tokens) -> Gemini 3 Pro via Bifrost - **Math/reasoning** -> DeepSeek R1 via Bifrost - **Private/offline** -> Local MLX via Bifrost (port 30080 routes to local MLX server) - **Code review consensus** -> Multi-model via PAL `consensus` - **Parallel multi-model research** -> PAL `clink` (when you need multiple model perspectives simultaneously) - **Architecture planning** -> Claude Opus native subagent (Plan mode or `Plan` subagent type) ## Route Selection | Task Type | Cloud Model | Local Model | Route | | --- | --- | --- | --- | | Research (single) | Gemini 3 Pro | mlx-community/Qwen3-235B-A22B-4bit | Bifrost | | Research (multi) | Multiple | mlx-community/Qwen3-235B-A22B-4bit¹ | PAL clink | | Complex Coding | Claude Opus | mlx-community/Qwen3.5-122B-A10B-4bit | native subagent | | Fast Tasks | Claude Sonnet | mlx-community/Qwen3.5-27B-4bit | Bifrost | | Code Review | Multi-model | mlx-community/Qwen3.5-27B-4bit | PAL consensus | | Architecture | Claude Opus | mlx-community/Qwen3-235B-A22B-4bit | native subagent | **Bifrost endpoint**: `http://localhost:30080/v1/chat/completions` (OpenAI-compatible) ¹ In local-only mode, `PAL clink` (multi-model) falls back to this single best local model. ## PAL MCP Tools (multi-model only) - **`clink`** - Paralle