Claude Code hooks that auto-switch model tier based on task complexity
-
Updated
Aug 2, 2026 - Python
Claude Code hooks that auto-switch model tier based on task complexity
Build a safer image gen product. Evaluate and tune your policies with gpt-oss-safeguard model on image prompts
Cut LLM costs 90%+. Route every prompt to the cheapest capable model. 6-level complexity score in 12 ms. Zero GPU.
Experimental transparent broker for the Claude Code CLI: classifies each prompt locally (deterministic, zero tokens, no AI calls) and applies the right reasoning effort level before inference same model, same session, exactly one forward, fail-open by design.
Smart LLM routing layer — classifies prompts and routes to the optimal model (Ollama, OpenAI, Anthropic, Groq) via Bifrost gateway. Cost-optimized, quality-first, and local-only strategies.
Add a description, image, and links to the prompt-classification topic page so that developers can more easily learn about it.
To associate your repository with the prompt-classification topic, visit your repo's landing page and select "manage topics."