gpd-explain

Invoke the skill with the concept name as an argument. It validates project context, searches local files for mentions of the concept, and spawns an explainer agent that writes a structured report to .gpd/explanations/ grounded in your current phase and manuscript.

After the explanation is written, a gpd-bibliographer subagent audits every reference in the literature guide, checking titles, authors, years, and URLs. Unverified or fabricated references are corrected, flagged, or removed, with the audit saved as a separate report.

Yes, the skill supports standalone mode when no project exists. You must provide an explicit concept or topic, and the explanation is generated from the request plus any relevant files in the current working directory.

It writes a structured markdown report to .gpd/explanations/{slug}-EXPLAIN.md containing an executive summary, prerequisites, formal equations, project connections, and a literature guide. A citation audit file {slug}-CITATION-AUDIT.md is also produced when references are checked.

The skill asks one focused clarifying question when the request is too vague and the active project context cannot disambiguate it. It returns a checkpoint message describing what disambiguation is needed and why it matters.