purity-auditlisted

# Purity Audit Audit whether code or docs read as first-written — no tracing residue, no decoration, no dead content, no doc/impl gap, no language/platform residue. Drives the 16-point purity-checklist against a target and produces an audit report. The purity principle's "No residual artifacts" rule made operational and reusable. ## When to use - A file, skill, doc set, or the whole pack reads like it's been edited many times — version tags in prose, "previously X now Y", dead links, decorative diagrams, "保留 / 已观测" labels. - Pre-ship cleanliness gate before a release or a doc refresh. - Enforcing "No residual artifacts" on a concrete target. **Not for:** syncing formal docs to code behavior / endpoints / TODOs (use `documentation-audit` — that's factual drift, this is cleanliness); bug-hunting a diff (use `code-review`); reducing code complexity (use `simplify`); machine-detectable lint (use `linting`); structural refactor (use `refactoring`). Dead-content removal *is* in scope; restructuring code is not. ## Steps ### 1. Scope the target + load the checklist Pick the target: a single file, a skill folder, a doc set, or the whole pack. Load [${CLAUDE_PLUGIN_ROOT}/references/purity-checklist.md](${CLAUDE_PLUGIN_ROOT}/references/purity-checklist.md) — the 16-point A–E grid is the audit criteria. ### 2. Grep-driven sweep — categories A / B / C These yield to batch grep. Run the checklist's "怎么查" commands: - **A (tracing residue)** — `（.*新` / `（.*保留` / `（.*事件` / `（.*已观测`