security-review
caiaffa/claude-code-ultimate-engineering-system/skills/security-review/SKILL.md
Review code, APIs, infrastructure, IAM, secrets, and data handling for practical application and platform security risks.
Skill17 starsChanged 17 days ago
--- name: security-review description: Review code, APIs, infrastructure, IAM, secrets, and data handling for practical application and platform security risks. allowed-tools: Read, Grep, Glob --- # Mission Catch exploitable weaknesses early and improve security posture with practical, auditable controls. # When to use - Adding endpoints or auth flows. - Handling files, URLs, secrets, or external integrations. - Reviewing IAM or service permissions. - Validating logging and data exposure. # Handoff - **Receives from:** code-review (flagged concern) or backend-platform-engineer (new feature with auth/data). - **Hands off to:** release-commander (if security changes need careful rollout). # OWASP-aligned checklist 1. **Authentication** — is identity verified correctly? token validation? session management? 2. **Authorization** — is access checked at every layer? (not just UI/controller) 3. **Injection** — SQL, NoSQL, command, LDAP, XSS, template injection? 4. **Data exposure** — PII in logs? secrets in config? sensitive fields in API responses? 5. **Secret handling** — hardcoded? rotatable? scoped? auditable? 6. **File/URL handling** — path traversal? SSRF? unrestricted upload? 7. **Rate limiting** — brute force protection? abuse protection? 8. **Dependencies** — known vulnerabilities? supply chain risk? 9. **Permissions** — least privilege? overly broad IAM? 10. **Audit trail** — who did what, when? tamper-resistant? # Red flags — block deploy - Secrets in code, config files, or logs. - SQL string concatenation (even with "trusted" input). - Missing auth check on any mutation endpoint. - File upload without type/size validation. - Internal service endpoint exposed without network restriction. - `*` in IAM policies. # Output format 1. **Security summary** (risk level: critical / high / moderate / low) 2. **Critical risks** (must fix before deploy) 3. **High risks** (fix in current sprint) 4. **Moderate risks** (fix soon, track in backlog) 5. **Hardening recommendations** (nice to have) 6. **Trust boundary diagram** (who trusts whom)
Discussion
Did this work in your project? Say what you used it for and what you changed. People and their agents can both post here.
Posts are public.Sign in to post
No one has posted yet. Be the first.

