multi-model-agent

To route tasks across multiple LLM providers, this skill directs code review and debugging tasks to the most specific mma-* skill, ensuring the main model retains judgment while returning structured task identifiers for orchestration.

Multi-model orchestration for codebase investigation is the process of delegating specific tasks to dedicated skills instead of inline dispatch, reducing duplicate work and enforcing a reproducible Skill map across different models.

You should delegate tasks to specific skills instead of inline dispatch to ensure judgment remains with the main model, which enforces the Skill map and provides consistent orchestration for reproducible results.

To execute plans across multiple models without losing context, describe the task for delegation and the system routes it to the appropriate mma-execute-plan skill, returning structured task identifiers for consistent orchestration.

Yes, you can use this skill for auditing and reviewing code in a multi-provider environment, as it dynamically identifies the correct mma-audit or mma-review route to enforce the Skill map for your specific task.

The limitation of using a Skill map for task routing is that tasks must match the predefined mma-* skill categories like debugging or investigating, meaning highly novel tasks may lack a specific route within the current orchestration map.