Setup & Installation
Or with the ClawHub CLI, for registry-managed skill folders outside a full OpenClaw workspace:
What This Skill Does
Routes a prompt to the best-performing LLM based on aggregated peer-reviewed council evaluations. Given a query, the API returns the model that has historically performed best on similar tasks, along with confidence and relevance scores. Returned model IDs match the OpenRouter catalogue, so they chain without remapping.
Instead of guessing or running manual benchmarks, it queries thousands of past peer-reviewed evaluations to surface the model that actually wins on similar tasks.
When to use it
- Picking the best model before answering a complex coding question
- Automating model selection in a multi-step AI pipeline
- Comparing ranked model performance for a specific task domain
- Routing user queries to the optimal LLM without manual benchmarking
- Chaining model selection with OpenRouter for automatic best-model dispatch
Example Workflow
Here's how your AI assistant might use this skill in practice.
User asks: Which LLM should I use to write a Python web scraper?
- 1Calls POST /v1/route with the query
- 2Reads the returned model field (e.g. anthropic/claude-sonnet-4)
- 3Checks confidence and relevance scores to assess recommendation strength
- 4Calls that model via OpenRouter using the returned model ID
- 5Returns the generated code from the best-matched model
Python web scraper code generated by the top-ranked model, selected based on 20 nearby council evaluations with 0.85 confidence
Requirements
Accounts, API keys, or tools you or your AI assistant may need to set up while using this skill.