openplan-garden — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited openplan-garden (Agent Skill) and scored it 96/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 1 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 1 flagged
The text {match} tells the agent to skip the normal "ask the user first" gate. Used adversarially it removes the human-in-the-loop check before destructive or sensitive actions, turning a normally-gated agent into a fire-and-forget executor.
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
Use this after $openplan-core for document quality, periodic documentation maintenance, and semantic drift inspection. Garden is about preserving the clarity of human goal/intent over time.
Garden first reports. It does not rewrite durable docs or memory unless the user explicitly asks for a cleanup/edit pass.
Use garden as the periodic documentation cleanup workflow. A periodic pass checks:
Garden first reports proposed cleanup. It may edit only when the user explicitly asks for a cleanup pass, and durable goal/intent changes still require $openplan-record and human confirmation through the human-facing coordinator.
references/checklist.md.references/report-contract.md.$openplan-record before editing durable memory/docs.Garden can operate on any docs root: .openplan/, docs/, repo root, a separate docs repo, or a path supplied by the user. Treat document-map paths as relative to the selected root. If the root is ambiguous, report the ambiguity before scanning broadly.
G0: misleading human goal/intent, wrong source of truth, or future agent likely to act incorrectly.G1: missing rationale, stale unmarked claim, authority/layer confusion, or broken decision boundary.G2: duplicated content, weak traceability, incomplete next action, or zero-context readability gap.G3: wording, formatting, title, or navigation improvement.Use deterministic tools only when available:
If not available, continue with semantic review and state checks not run.
Use assets/claude-code-command.md as a slash command or prompt seed.
~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.