api-exploit-prover

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The prompt explicitly demands "exact request and response pairs" and "exact request signatures (method, path, key headers, payload hash)" and per-attempt auth context, which implies including Authorization headers/cookies/API tokens verbatim in outputs and thus requires handling secrets directly.

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.80). The skill's Execution Workflow (Phase 1: "Replay original request as control") and Evidence Requirements ("Exact request and response pairs") explicitly require the agent to fetch and ingest responses from the provided target_base_url (arbitrary/untrusted API endpoints), so untrusted third‑party content could be read and influence subsequent actions.