openmind-glossarylisted

Resolve a term against Open Mind's deterministic, source-traceable glossary — verbatim definitions with file:line provenance, exact-token lookup, honest "not found" for unknown terms. Runs Open Mind's real Python implementation through the skill bridge.
HelloThisWorld/agent-skill-verification-template · ★ 1 · AI & Automation · score 70

Install: claude install-skill HelloThisWorld/agent-skill-verification-template

# openmind-glossary This skill is the eval-harness face of Open Mind's `glossary` capability (https://github.com/HelloThisWorld/open-mind, `skills/glossary/SKILL.md`). The answers are produced by Open Mind's actual extraction + lookup code (`openmind/glossary.py`, `openmind/structure.py::term_usage`), reached through `openmind/skill_bridge.py`; nothing in this repo reimplements them. ## Behavior under contract - **Verbatim, never generated** — the definition is the exact original text lifted from the corpus's authoritative source (dedicated glossary file, definition table/line, README acronym expansion, doc-comment). - **Provenance mandatory** — every entry carries `source_file:line_number`; the claim cites that exact line, and the validators re-read it on every run. - **Exact-token lookup** — no similarity, no paraphrase. Unknown terms return `insufficient_evidence` with zero claims. - **Grounded usage profile** — when the term is also a code symbol, a second claim cites its real definition site from the deterministic structure map. ## Question forms `define <term>` · `glossary <term>` · `what is <term>`