The Agent Skills Directory

[PROMPT_INJECTION]: Indirect Prompt Injection Surface. The skill facilitates interaction with untrusted external data from Facebook, which could potentially contain malicious instructions designed to influence agent behavior.
Ingestion points: The skill fetches data from Facebook pages, posts, comments, and insights via the rube MCP server.
Boundary markers: Absent. The skill does not provide specific instructions or delimiters to help the agent distinguish between its own operational instructions and the data retrieved from external sources.
Capability inventory: The skill includes significant write capabilities, such as creating and scheduling posts, replying to comments, and managing ad accounts.
Sanitization: Absent. No sanitization or validation logic is defined for the external content processed by the skill.
[NO_CODE]: This skill consists purely of configuration metadata and natural language instructions. It does not ship with any scripts, binaries, or executable code, reducing its direct attack surface.
[SAFE]: No malicious patterns, such as hardcoded credentials, obfuscated code, or unauthorized exfiltration attempts, were detected. The use of external resources is consistent with the skill's primary purpose of Facebook automation.

facebook-automation