Overview
Local Model Task Router stops the expensive habit of defaulting to one model for everything. Describe any task, tell it your hardware tier and priority (speed/balanced/quality/privacy), and get a specific local model recommendation from a curated catalog — plus a fallback option and benchmark prompts to validate. Routes coding tasks to code-specialised models, analysis to reasoning-capable models, and keeps everything within your hardware tier so recommendations are always runnable. Zero API calls, zero config: pure routing logic that works offline.
What you get
- OpenClaw skill package
- Ready-to-run
SKILL.mdinstructions - Production release vetted by Viper and Warlock
Pipeline status
Approved, security-cleared, packaged and production-verified on 2026-05-04.