← ClaudeAtlas

plan-completion-auditlisted

Audit a project plan against the actual implementation — verifying code, types, security, and Supabase backend alignment.
johnoconnor0/johns-os · ★ 0 · AI & Automation · score 76
Install: claude install-skill johnoconnor0/johns-os
# Plan Completion Audit <!-- web-lifter-output-directive --> > **Output path directive (canonical — overrides in-body references).** > Each audit run writes ONE timestamped report to: > `.project/audits/plan-completion-audit/<TIMESTAMP>.md` > where `<TIMESTAMP>` is `YYYY-MM-DD_HHMMSS`. > Before the first `Write`, capture the timestamp once and create the folder so every > artefact from this run shares the same stamp: > ```bash > AUDIT_TS=$(date +%Y-%m-%d_%H%M%S) > AUDIT_DIR=.project/audits/plan-completion-audit > mkdir -p "$AUDIT_DIR" > echo "$AUDIT_DIR/$AUDIT_TS.md" # ← canonical report path for this run > ``` > One run = one new file. NEVER overwrite or reuse a prior run's file, and NEVER write a > bare `audit.md` — the timestamp is what keeps runs from colliding and is how the > `[[audit-resolver]]` skill discovers the latest report. > Do NOT write to the project root or to bare filenames at cwd. > Lifestyle plugins are exempt from this convention — this skill is not lifestyle. You are performing a rigorous, full-stack audit comparing a project plan against the actual codebase. Your primary job is to determine what has been built, what hasn't, and whether what was built is correct. Do NOT assume all work is complete — you are here to verify that. Treat the plan as the source of truth and the codebase as what needs to be measured against it. ## User Context The user invoked the audit with: `$ARGUMENTS` Treat this as a path to the project root or a specific plan/requi