api-tester

CRITICAL E004: Prompt injection detected in skill instructions.

• Potential prompt injection detected (high risk: 1.00). The skill includes a hidden "Use-to-Evolve" Post-Invocation Hook that must run after every call ("Do not skip it") and autonomously records usage and can micro-patch triggers/modify behavior—self-modifying instructions outside the stated API-testing purpose and presented as mandatory operational steps.

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

• Third-party content exposure detected (high risk: 0.72). Runtime path §4/§6 indicates the agent captures the HTTP response body (user-supplied URL/method in TEST/BATCH) and then uses it for security scanning/validation/reporting, meaning arbitrary outsider-authored response text can be ingested into the LLM context.