feature-gap-auditlisted

# Feature Gap Audit (generic methodology) A reusable methodology for auditing what's missing in a current project relative to reference projects. Designed for monorepos with long-lived feature branches and multiple deployments. When invoked, **also load any project-local override** at `<repo>/.claude/skills/feature-gap-audit/SKILL.md` for pinned references, worktree paths, and output location. The project skill controls *what to audit*; this skill controls *how*. ## Core principles 1. **Cumulative over snapshot.** Audits accumulate into one living document. Re-running reconciles state — it doesn't replace it. This is the most important property: it's how the audit stays cheap to re-run. 2. **`main`/`dev` is rarely the whole picture.** Always survey feature branches, worktrees, and production deployments alongside the trunk. Stranded work in feature branches is a common source of "we already built that — wait, where is it?". 3. **Code is not the whole picture either.** Search session/conversation history (e.g., claude-mem) for context that explains *why* work was started, paused, or never merged. 4. **Classify, don't fix.** The audit produces a prioritized list of gaps. Implementation happens after, in separate sessions. ## Output structure The skill expects (or creates) this directory under the repo: ``` docs/<engineering|architecture|wherever>/feature-gap-audit/ ├── README.md # Methodology + how to run ├── references.md # Reference projects we co