routing-matrixlisted

# Routing Matrix — Model Dispatch Configuration This skill defines a routing matrix: which AI models are assigned to which agent roles, how fallback chains resolve, and what policies govern model selection at dispatch time. ## What this is A structured YAML configuration that maps model pools (cloud, local, direct API, MLX) onto agent roles (orchestrator, advisor, builder, qa, linter, kaizen, logger) and declares the policies that shape dispatch behavior. **This skill ships as a template.** The role structure and governance policies are pre-configured. You fill in model names, provider lanes, and ban lists. ## Quick start 1. Copy `templates/routing-matrix.template.yaml` to your project root. 2. Rename it to `routing-matrix.yaml`. 3. Fill in the `<...>` placeholders with your model names, providers, and context windows. 4. Run the schema validator: `python3 routing-matrix/scripts/validate.py routing-matrix.yaml` ## Role structure The matrix pre-defines 10 roles in a fixed sequence. Each role has: | Field | Description | |---|---| | `primary` | Model ID to use when available | | `provider` | Provider lane (openai, anthropic, ollama-local, ollama-cloud, openrouter, zai, mlx-local, etc.) | | `strength` | Human-readable description of why this model fits this role | | `sequence_order` | Position in the sequential pipeline (1 = first to run after orchestrator) | | `fallback_chain` | Ordered list of `provider/model-id` alternatives if primary is unavailable | | `lens` | Wha