namht-security-auditlisted

# namht-security-audit — whole-repo security sweep A repo-wide security audit (broader than `/namht-review`, which is per-file). It enumerates the **attack surface**, checks it against a security checklist, and reports prioritized findings with fixes. Read-only — it does NOT change code (hand fixes to `/namht-fix-bug` or `/namht-build`). ## Inputs & grounding - **Checklist:** `knowledge-base/review-skills.md` §2 SECURITY + §8 AI/LLM (fallback bundled `references/review-skills-universal.md`). Plus project rules (Section 14) if present. - **Attack surface (KB + code):** find every external entry point — HTTP/gRPC routes, message consumers, scheduled jobs, CLI, file/upload handlers, auth flows — via KB `03-entry-points.md` / `09-auth-security.md` / `11-api-docs.md` and by reading the controllers/handlers. These are where untrusted input enters. - If no KB, fall back to Grep/Glob (note reduced coverage). ## Audit categories (cover each; cite file·function·line) 1. **Input validation** — every entry point validates untrusted input at the boundary? whitelist > blacklist? 2. **Injection** — SQL/NoSQL (string-built queries), command, XSS, path traversal, SSRF, template injection. 3. **AuthN / AuthZ** — auth on every protected endpoint; **IDOR** (can a user reach others' resources?); role/permission checks at the top; tenant isolation; token expiry/rotation. 4. **Secrets & crypto** — hardcoded secrets/keys, secrets in logs, weak hashing (MD5/SHA1/plain), insecure rand