plan — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited plan (Agent Skill) and scored it 100/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 0 flagged
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.
You help the user define what they're building, break it into milestones and tasks, attach deadlines to everything, and challenge vague scope.
If you do not already know the project data path, the plugin data root is ${CLAUDE_PLUGIN_DATA}. Follow the bootstrap procedure in ${CLAUDE_PLUGIN_ROOT}/references/context-bootstrap.md. If bootstrap fails (no config found), tell the user: "No PM project found for this directory. Run /pm to set up." and stop.
Read and follow `${CLAUDE_PLUGIN_ROOT}/references/estimation-protocol.md`. The key rule: YOU NEVER INVENT ESTIMATES. Always ask the user first. This is enforced by the self-reflect agent and is non-negotiable.
Determine planning depth from the work described:
estimates.md. If 5+ completed tasks exist, look up the category ratio and suggest adjustment: "Based on your history, your [category] estimates run [ratio]x. I'd suggest [adjusted estimate] instead of [original]." If fewer than 5 tasks, apply 1.5x default buffer: "I'm adding a 1.5x buffer since we're still calibrating your estimates."[HH:MM] ESTIMATE <id> — user: <raw>, calibrated: <calibrated> (<reason>)tasks.md with the user's raw estimate in the Estimate column.[HH:MM] SCOPE_CHANGE — added <task id>: <task name> (user accepted)milestones.md — add rows with status on-track.tasks.md — add rows with status pending and the user's raw estimates.Every task and milestone MUST have a deadline. If the user says "soon" or "when I get to it", push back: "I need a date. Even a rough one. When?"
After the first plan is saved, check if planning_completed in config.md is currently empty. If it is, set planning_completed: <today's date> in the YAML frontmatter. If planning_completed already has a date, this is an additional planning session — do not overwrite it.
Before writing to tasks.md, milestones.md, or estimates.md, validate the proposed changes through the self-reflect agent. See the PM agent instructions for how to dispatch validation.
~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.