LewLM

Evaluate or design AI tooling for VS Code Copilot: custom agents, skills, instructions, hooks, MCP servers, plugins, prompts, model routing, and workflow automation. Use when setting up or improving an AI-assisted development environment.

Plan and run validation for LewLM changes. Use for pytest selection, targeted test mapping, Python hook/script syntax checks, docs-only checks, long_running test decisions, and pre-PR confidence.

Review LewLM local AI runtime, routing, serving, MLX, llama.cpp, OpenAI-compatible adapter, benchmark, and fallback changes. Use for local-first runtime correctness and capability reporting.