| name | security-review |
| version | 0.1.0 |
| description | Security-focused review rules for command and secret-sensitive changes. |
| priority | 90 |
| budget-multiplier | 1 |
| allow-fallback | true |
| incompatible-with | ["fast-mode"] |
security-review Skill
Purpose
Use this skill when a task touches command execution, package scripts, CI, deploy files, authentication, secrets, receipts, memory, or user-controlled paths.
Read First
If present, review these before making security-sensitive edits:
- SECURITY.md
- package.json
- .github/workflows/
- src/security-policy.ts
- src/commands/verify.ts
- any file that defines command execution, receipts, secrets, auth, or writable paths
Rules
- Treat install scripts, lifecycle hooks, shell commands, workflows, Docker files, and deploy files as high-risk surfaces.
- Never expose secret values in logs, receipts, docs, tests, or examples.
- Prefer allowlists and structured parsing over broad string matching for security decisions.
- Do not weaken SWD verification, receipt integrity, dry-run behavior, or command review prompts.
- When a change introduces a new writable path, explain how it is constrained.
Verification
- Check that sensitive outputs are redacted before storage.
- Check that command-affecting changes still require explicit human confirmation.
- Prefer read-only verification for CI-facing checks.
Output
When finished, summarize:
- what security-sensitive files changed
- what risks were checked
- what verification was run
- any remaining assumptions or unverified areas
