functional-area-resolver

Compress large AGENTS.md routing files by replacing per-row skill entries with functional-area dispatchers. This groups sub-skills under area labels with a (dispatcher for: ...) clause, allowing the LLM to route correctly while drastically reducing file size.

Functional-area dispatcher routing is a two-layer compression technique that replaces long routing tables with grouped area labels. It funnels requests to sub-skills via a (dispatcher for: ...) list without listing every individual row, preserving sub-skill reach while shrinking context footprint.

Routing table compression requires two preconditions: the routing file must exceed a specific size threshold, and the working tree must be clean. These gates ensure the compression is necessary and prevent uncommitted changes from interfering with the generated routing fragment.

Validate routing accuracy after compression by running gating verification checks. The workflow requires passing both structural routing-eval and harness AB tests before committing the dispatcher-based routing fragment to ensure sub-skill precision is preserved.

You should not use area-based dispatchers if your routing file does not exceed the required size threshold or if your working tree has uncommitted changes. The compression is designed specifically for large files where routing tables exceed context budgets in Brain or GBrain deployments.