aped-glossary

To maintain a consistent domain glossary, you can use a corpus-driven extraction tool to scan existing PRD, architecture, and story artifacts for new or drifted terminology. The tool proposes canonical names and synonyms, appending only user-approved revisions to a shared glossary file to enforce naming consistency.

Terminology drift occurs when teams use inconsistent naming for the same concepts across PRDs, architecture documents, and user stories. This drift causes confusion and wasted tokens downstream. A canonical glossary eliminates drift by recording explicit canonical names and listing synonyms under an avoidance section.

You can extract candidate terms by running a glossary builder that discovers upstream corpus sources from your APED artifacts, such as product briefs, project context, and stories. The tool identifies NEW, DRIFT, and STALE term candidates, then requires your per-term confirmation before writing any revisions.

No, the glossary builder operates under strict non-invention rules. It only extracts candidate terminology from your existing APED corpus and requires explicit user confirmation for each term before appending any changes, ensuring the canonical vocabulary strictly reflects your actual project artifacts.

Yes, you can audit naming consistency in a Claude Code workflow by running a glossary skill that reads your APED config.yaml, discovers corpus sources, and signals terminology drift. It records synonyms to avoid and supports downstream drift detection against the maintained glossary file.

To canonicalize project terminology safely, the glossary tool preserves prior entries and only changes what you explicitly approve. It appends revisions to an output glossary file after collecting per-term confirmations, ensuring your historical domain vocabulary remains intact while new drift is corrected.