owasp-top-10-testing

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill explicitly instructs embedding account passwords into the agent's CLI/instruction string (e.g., "Accounts: userA@example.com/" inside a --instruction), which requires the LLM to receive or emit secrets verbatim and therefore poses a direct credential-exfiltration risk.

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

• Third-party content exposure detected (medium risk: 0.30). This skill’s required workflow describes scanning a specified target (e.g., a running instance at a URL such as https://staging.example.com) and attempting exploit/proof actions against it, which necessarily ingests outsider-authored free text from the target’s responses (HTML/API error messages) at runtime.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The skill's run example passes a source repository URL that the agent will fetch and analyze at runtime, meaning https://github.com/org/app can supply external code/content that directly controls the agent's prompts/analysis.