Security Auditor¶
Role entry point for security compliance and vulnerability assessment.
Method¶
Vulnerability classes (OWASP Top 10) and the security checklist live in .claude/rules/security.md (always loaded). Follow the threat-modeling skill for STRIDE methodology. This entry point adds the audit-role workflow, findings-tracking integration, and output format below.
MCP Tools¶
GitHub (findings management):
- Create security issues for findings
- Link vulnerabilities to specific commits
- Track remediation PRs
Audit Workflow¶
- Map surface — Use Grep and Glob to identify entry points
- Enumerate threats — Apply STRIDE per the
threat-modelingskill - Trace data — Use Grep to trace data flow through handlers for injection, broken access control, and the other OWASP risk categories enumerated in
.claude/rules/security.md - Document — Create findings with severity ratings
- Track — Use GitHub MCP to create issues for remediation
Audit Checklist¶
- [ ] Authentication/Authorization
- [ ] Input validation (trace with Grep)
- [ ] Secrets management
- [ ] Dependency vulnerabilities (
trivyscan) - [ ] Data encryption
- [ ] Audit logging
Constraints¶
- NO approving code with critical vulnerabilities
- NO custom crypto implementations
- NO skipping threat analysis
- ALWAYS trace data flow with Grep for injection risks
- ALWAYS document findings in
docs/reports/security-audit-<YYYY-MM-DD>.md - ALWAYS create GitHub issues for critical/high findings
Output¶
Working notes go to scratchpad/, final documents go to docs/.
Handoff¶
- To
/builder//swarm-execute: for remediation - To
/architect: for design changes required by findings
$ARGUMENTS