Garry's Opinionated OpenClaw/Hermes Agent Brain
TypeScript
{"intent": "let's do EIIRP on what we just built", "expected_skill": "eiirp"}
{"intent": "time for EIIRP — wrap this all up", "expected_skill": "eiirp"}
{"intent": "I want everything in its right place after today", "expected_skill": "eiirp"}
{"intent": "let's get everything in its right place before EOD", "expected_skill": "eiirp"}
{"intent": "make this re-doable for next quarter", "expected_skill": "eiirp"}
{"intent": "let's DRY this up across our skills", "expected_skill": "eiirp"}
{"intent": "please file all of this properly", "expected_skill": "eiirp"}
{"intent": "organize all of this work so it's findable later", "expected_skill": "eiirp"}
{"intent": "archive this research thread once we're done", "expected_skill": "eiirp", "ambiguous_with": ["data-research"]}
// Routing-eval additions for skills/eiirp v1.1.0 (auto-fire gate merge from
// deep-analysis-brain-auto@fc834ee). Merge into skills/eiirp/routing-eval.jsonl
// once the RESOLVER.md eiirp row carries the new trigger phrases
// ("analyze this document", "deep analysis", "review this report").
// The gate itself is ALWAYS-ON (fires at delivery time, not via routing);
// these fixtures cover the explicit-ask surface only.
{"intent": "run a deep analysis on this diligence packet from acme-example and make sure it's findable later", "expected_skill": "eiirp"}
{"intent": "can you analyze this document I just uploaded and file the takeaways somewhere permanent?", "expected_skill": "eiirp"}
{"intent": "here's a PDF — deep analysis please, then save to brain", "expected_skill": "eiirp", "ambiguous_with": ["capture"]}
{"intent": "please review this report on the widget-co pilot and give me structured findings", "expected_skill": "eiirp"}
{"intent": "what page is the indemnity clause on in this contract?", "expected_skill": null, "ambiguous_with": []}