| name | security-review |
| description | Audit code changes for security vulnerabilities before merging. Reports findings; never auto-fixes silently. |
Security Review
A focused checklist for catching real vulnerabilities in a change set. Apply to the diff first, then widen only if a finding points elsewhere.
Threat checklist
- Injection — untrusted input flowing into SQL, shell, OS commands, eval, template engines. Look for string concatenation instead of parameterized queries/argument arrays.
- XSS — user data rendered into HTML/JS/DOM without escaping. Check
innerHTML, dangerouslySetInnerHTML, unescaped template interpolation.
- AuthN / AuthZ — missing authentication, missing ownership/permission checks, IDOR, privilege escalation, trusting client-supplied roles.
- Secrets — hardcoded API keys, tokens, passwords, private keys, connection strings; secrets logged or committed. Should come from env/secret store.
- Path traversal / file access — user-controlled paths joined without normalization/allow-listing.
- SSRF — server-side requests to user-supplied URL/host without allow-listing.
- Deserialization / parsing — untrusted data fed to unsafe deserializers.
- Crypto — weak algorithms (MD5/SHA1 for passwords), hardcoded IVs, ECB mode, missing TLS verification, predictable randomness.
- Sensitive data exposure — PII/secrets in logs, error messages, or API responses; verbose stack traces leaked to clients.
- Dependencies — newly added packages: reputable, pinned, free of known CVEs.
- Resource & DoS — unbounded loops/request sizes, missing timeouts, ReDoS.
- Race conditions / TOCTOU — check-then-act without locking.
Method
- Identify the trust boundary: where does untrusted input enter, and where does it reach a sink (DB, shell, filesystem, network, HTML)?
- Trace each tainted value from source to sink; a vuln exists when it reaches a dangerous sink without validation/encoding/parameterization.
- Prefer allow-lists over deny-lists; prefer library-provided escaping over manual sanitization.
- Only flag issues with a concrete exploit path — no speculative noise.
Report format
For each finding:
[severity: high|medium|low] <title>
location: path/to/file.ext:LINE
issue: <what is wrong and the input that triggers it>
impact: <what an attacker gains>
fix: <the minimal concrete remediation>
Order findings by severity. If you reviewed and found nothing actionable, say so explicitly.
