functional-area-resolver

To compress agent routing, convert granular skill-per-row tables in RESOLVER.md or AGENTS.md into functional-area dispatchers. This shrinks routing files while preserving intent-to-skill routing accuracy through a two-layer structure.

Functional-area dispatch compression is a routing optimization technique that consolidates long skill-per-row tables into one entry per functional area. It uses a "(dispatcher for: ...)" clause to maintain correct sub-skill selection while reducing file size.

After compressing AGENTS.md, verify routing accuracy by running routing-eval structural checks and an LLM harness. This evaluation-focused safety step ensures the compressed dispatcher preserves the original intent routing before committing.

Yes, functional-area dispatchers are designed for large routing files hitting context limits. The compression pattern can reduce a ~25KB RESOLVER.md to ~13KB while maintaining reliable, correct intent-to-skill routing.

Agent routing compression requires YAML frontmatter-driven discovery plus area entries with a "(dispatcher for: ...)" list. This format allows the model to drill into sub-skill options while keeping the top-level routing table compact.

You should avoid functional-area routing compression if your RESOLVER.md or AGENTS.md is small enough to fit comfortably within context limits, as the two-layer routing structure adds dispatch overhead that is unnecessary for compact agent routing files.