aiagent: Skill for Claude Code

Security Review

Security review for:

Instructions

You are orchestrating a Security Review cycle for the aiagent project.

Phase 1: Security Analysis

1. Use the Agent tool with subagent_type: "security" to analyze the code
2. Pass it the feature description and list of changed files (if known)
3. The security agent will:
   • Run grep-based scans for command injection, path traversal, API key leakage, MongoDB injection patterns
   • Review each changed file against the project's threat model
   • Propose specific remediations for any issues found

Phase 2: Review 4. Use the Agent tool with subagent_type: "security-reviewer" to review the security analysis 5. Pass the security agent's full output to the reviewer 6. The reviewer will independently run the automated security scans and score against the 6 pillars

Phase 3: Iterate 7. If overall score >= 8/10 (APPROVED): Report the security posture and scores 8. If overall score < 8/10 (REVISE):

• Extract Critical Issues from the reviewer
• Send issues back to the security agent for remediation proposals
• Re-review with the security-reviewer
• Repeat until APPROVED or 3 iterations maximum

Phase 4: Report 9. Present:

• Security findings summary
• Remediations applied or recommended
• Final pillar scores table
• Any remaining warnings

Important

• Maximum 3 review iterations
• The security agent PROPOSES FIXES; the reviewer only READS and SCORES
• Critical issues (injection vulnerabilities, key leakage) block shipping
• Warnings are non-blocking but should be tracked