audit

Files (3)

• SKILL.md
• findings-menu.md
• project-mode.md

SKILL.md (first lines)

# /audit

On-demand audits. Three distinct targets, no depth menu, no recency
recommendation. Routine doc-drift is handled automatically by
`/session-end` — run `/audit` only when you have a specific
concern that warrants a deeper look.

## Project Amendments

At the start of every invocation, check whether
`docs/rpm/skills/audit.md` exists in the consuming project. If it
does, read it and apply its contents as additional project-specific
instructions for this skill. Amendments may add audit dimensions,
require extra scans, or extend the report format. They cannot remove
or override plugin defaults — on conflict, this SKILL.md wins.

## Routing

Parse `$ARGUMENTS`:

- `quick` → run the **Quick** audit below (mechanical scan.sh only,
  no LLM scan)
- `documents` (or `docs`) → run the **Documents** audit below
- `project` → run the **Project** audit described in
  [project-mode.md](project-mode.md). Read that file and follow it
  end-to-end — it contains the full 5-phase protocol, the analyze
  dimensions, and the deliverables spec.
- empty or unrecognized → print the usage block and stop:

  ```
  ## /audit — pick a target

  - `/audit quick` — mechanical scan.sh only. Git state,
    agent instructions size, NOT_IMPLEMENTED, broken refs,
    daily-log gap, session marker, spec inventory drift,
    log/tracker staleness.
    Zero LLM tokens for the scan itself. ~5sec.
  - `/audit documents` — scan docs + agent instructions + memory +
    session drift via the rpm:auditor review agent. Scored findings,
    codify repeat offenders. ~3min.
  - `/audit project` — full consultant review: code, architecture,
    inward + outward research, 7-dimension analysis, saved plan file.
    ~30min+.

  Routine doc-drift runs automatically at /session-end.
  ```

---

## Target: Quick

Mechanical drift scan only — no `rpm:auditor` subagent, no LLM
scan, no 110-doc walk. Runs `skills/session-end/scripts/scan.sh`
via direct shell injection and interprets the output.

Use when you want a fast "anything broken right now?" check
between session-ends, or to verify a fix landed, without committing
to a full documents audit.

### Phase 1: Scan (auto-injected, no tool call needed)

The `scan.sh` output below was produced by the shell before this
skill body reached you. Its results are already in this message —
do NOT re-run these checks as tool calls.

!`bash "${CLAUDE_PLUGIN_ROOT}/skills/session-end/scripts/scan.sh"`

Interpret the sections exactly as `skills/session-end/SKILL.md`
Phase 1a describes — the interpretation rules are identical.

### Phase 2: Assemble findings

From the scan output, collect actionable items into a findings list:

- `git` — only flag if the user seems unaware of uncommitted work
- `agent_instructions` — flag if `status=warn` or `status=critical`
  (fall back to legacy `claude_md` only if `agent_instructions` is
  absent from older scan output)
- `not_implemented` — flag only real source stubs, suppress meta
- `broken_refs` — always flag if `count > 0`
- `daily_log` — flag if `today_exists=false` AND `commits_since > 0`
- `session_marker` — informational, not a finding
- `specs_inventory` — flag if `unlisted > 0`
- `pm_docs_staleness` — flag if any `days > 3`
- `status_marker` — flag if `stale=true` (status.md's `**Last updated**`
  marker is older than the newest dated entry in the file; the mtime-based
  `pm_docs_staleness` check misses this internal-marker drift)
- `migration` — if `count > 0`, auto-migrate before presenting
  findings: `mkdir -p` target dirs, `mv` each `move=old→new` pair,
  `git add` both old and new paths. This is safe and non-destructive
  (git tracks the rename). Print a brief summary of what was moved.

### Phase 3: Present findings menu

Present via the **Shared Findings Menu** format described in
[findings-menu.md](findings-menu.md). Read that file for the exact
format, reply grammar, and execute flow.

Score each finding with `severity (0-40) + evidence (0-30) +
fix clarity (0-30)`. For scan.sh findings, evidence is generally
"single clear source" (20) since the scan is mechanical.

If no findings, print:

```
## /audit quick — no drift detected
```

and stop. Do NOT append to `past/log.md` for clean runs — quick mode
is meant to be cheap and invisible when nothing's wrong.

### Phase 4: Log results (only if findings were present)

Append a one-line run marker to `docs/rpm/past/log.md` Audit History:
`- YYYY-MM-DD — audit quick — N findings, M fixed, K skipped`

No findings detail beneath — keep quick mode's log footprint
minimal. If the user wants detail, they can re-run or upgrade to
`/audit documents`.

From dppdppd/rpm, MIT.

More from dppdppd/rpm