The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data from LinkedIn (posts, comments, headlines, and profiles) which may contain malicious instructions designed to influence the agent's behavior or output.
Ingestion points: Untrusted content enters the agent context through lib.ApifyClient.fetch_user_recent_comments and fetch_post_comments as described in the Steps section of SKILL.md.
Boundary markers: The skill includes a dedicated 'Untrusted content' section in SKILL.md that explicitly instructs the agent to treat fetched content as data, not instructions. It also references an external policy at ../../references/untrusted-content.md.
Capability inventory: The skill has the ability to read external data and trigger follow-up actions such as drafting replies via the linkedin-reply-handler skill.
Sanitization: The skill provides explicit rules to prevent fetched text from modifying draft bodies, adding unauthorized links, or bypassing user approval.

linkedin-thread-monitor