Attack Surface & Payload Generation Target: SKILL.md (skill) Attack surface (static, from the artifact's own text): Data access: credential Governance terms: no governance vocabulary present Modal statements: 0 (shape only — not counted as defenses) Surfaces mapped: 3 (social_engineering, data_exfiltration, instruction_override) - Data access: credential Social engineering using the target's own data domain - Sensitive data: credential Exfiltrate sensitive data via compliance framing - Instruction override (no governance vocabulary present) Direct instruction override; any governance named here is a target, not a defence Payloads generated: 4 Payloads executed: 0 Duration: ms Resilience: NOT MEASURED No agent was executed — no agent execution backend is wired to this engine, so no payload was run. Resilience is a property of a run, so this command reached no verdict on it. The payloads above are generated, not results. To run them against your agent yourself: hackmyagent red-team SKILL.md --json Payload text is under .results[].payloadInput