Pass
Audited by Gen Agent Trust Hub on Jul 28, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill provides the ability to read existing Facebook Page posts, which introduces a surface for indirect prompt injection.
- Ingestion points: Untrusted data enters the agent context via the
curlcommand targetinghttps://graph.facebook.com/v21.0/$PAGE/postswhen reading recent posts. - Boundary markers: The skill instructions do not specify the use of delimiters or clear separation between system instructions and retrieved post content.
- Capability inventory: The agent possesses the
Bashtool, allowing it to perform network operations (curl) and publish new content to Facebook, which could be abused if the agent follows instructions hidden within a retrieved post. - Sanitization: There is no evidence of message content sanitization or filtering before the agent processes the output of the post retrieval command.
Audit Metadata