facebook-profile

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill explicitly allows passing the FREVANA_TOKEN via a --token command-line argument (in addition to env vars), which requires the LLM to handle or emit a secret value verbatim in commands or outputs, creating an exfiltration risk.

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

• Third-party content exposure detected (high risk: 0.75). The required workflow calls an external Frevana API at runtime (via curl), then reads the returned response body as text and prints it to stdout (scripts/get_facebook_profile.sh:52-64), so any free-form content in that outsider-authored response can be fed into the agent’s LLM context.