| name | security-review |
| description | Load when reviewing Mandate402 for security risks across API, auth, database, payment, x402, worker, Morph, contract, or release boundaries. |
| metadata | {"compatibility":"OpenAI/Codex, Claude, Cursor, Gemini, and agents that can read repo-local Markdown skills","source":"repo-local bridge","upstream_note":"The requested sickn33 security-review slug is not exposed by the current upstream skill index; use cc-skill-security-review plus Mandate402 security skills."} |
Security Review
This is the repo-local security review router for Mandate402. It keeps the public trigger stable while routing agents to the narrower project and OSS skills that match the actual risk boundary.
Procedure
- Classify the boundary before reading code: API/auth, database, x402/payment, worker/reconciliation, Morph/contract, frontend trust boundary, or release/CI.
- Load the mandatory Mandate402 skill for that boundary first. Use
mandate402-runtime-security for API/auth/payment runtime work and mandate402-blockchain for Morph or contract-facing work.
- Load
api-security-hardening for public API hardening, cc-skill-security-review for general secure-code checklist coverage, and software-crypto-web3 for onchain/offchain risk.
- Verify with deterministic checks before reporting: repo-safety, focused tests, typecheck, contract tests, or route-level tests as appropriate.
- Do not mark a finding resolved unless code, tests, and docs agree.
Gotchas
- Do not treat client-side checks as authorization. Server routes and worker surfaces own enforcement.
- Do not log secrets, private keys, signed payloads, raw HMAC credentials, or local machine paths.
- Do not call unresolved payment truth final.
execution_unknown remains unresolved until correlation proves final truth.
- Do not rely on LLM-only security judgment for contracts or signing. Reproduce findings with tests or deterministic tools.
