plan-security-audit — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited plan-security-audit (Agent Skill) and scored it 45/100 (orange). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 1 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 1 flagged
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
*.sig, SIGNATURES) outside the documentation.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.
Role: Senior application security engineer.
Task: Exhaustive vulnerability audit (frontend, backend, auth, Supabase, deps, secrets), mapped to OWASP Top 10, then remediation plan. Audit & plan only — no code changes, no destructive testing.
| Skill | Does |
|---|---|
| plan-security-audit (this) | Plan with OWASP + Supabase-first burndown |
audit-security | Static security review (may fix) |
test-red-team | Adversarial runtime testing |
audit-db-schema | Schema health including RLS review |
Loop: see docs/PLAN-LOOPS.md — after plan-test-coverage, parallel with plan-perf-audit, before plan-docs-sync
Read references/preservation-contract.md. Acknowledge in output #1.
Safety guardrails:
| File | Contents |
|---|---|
references/owasp-supabase-scope.md | OWASP map, RLS-first methodology, scans |
references/output-templates.md | Burndown, phased hardening, secrets list |
1. Stack fingerprint
2. RLS-first pass (Supabase MCP if available)
3. Bundle/secret scan (no value echo)
4. Auth-path trace
5. Dependency CVE scan
6. OWASP remainder (injection, headers, misconfig)
7. Burndown + phased hardening plan
8. Guardrails + research citationsSentry MCP: security-related production errors. Firecrawl: current OWASP + Supabase patterns.
Enumerate every public table:
WITH CHECK on writes?#1 critical class: table without RLS + anon key in client = full data exposure.
Detail: references/owasp-supabase-scope.md
Inspect client bundle + source for:
service_role key (Critical — full RLS bypass).env committed / not gitignoredReport: file:line + type only — never the value.
npm audit, lockfile versions vs CVE databasesMap each finding to OWASP category.
Template: references/output-templates.md
Phases: Critical → High → Med/Low. Each remediation + "what must keep working".
Re-scan proposed after fixes (second pass).
[NEEDS VERIFICATION] + secrets-to-rotate (locations only)[NEEDS VERIFICATION].~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.