The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides functional shell commands using curl to interact with public APIs for social listening. Evidence found in references/listening.md includes recipes for fetching data from Reddit (reddit.com/.../new.json), Hacker News (hn.algolia.com), and Bluesky (public.api.bsky.app). These commands are well-documented and intended for data retrieval.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted data from external sources which could contain malicious instructions meant to influence the agent's behavior.
Ingestion points: External data enters the agent context via curl fetches from Reddit, Hacker News, and Bluesky, as well as browser-driven extraction from LinkedIn and X (references/listening.md).
Boundary markers: There are no explicit instructions or delimiters used to separate the external content from the system instructions, though a human-in-the-loop verification step is recommended before posting replies.
Capability inventory: The skill uses subprocess calls (curl) and generates output (comment drafts) that could be influenced by external input.
Sanitization: No explicit sanitization or filtering of the fetched content is performed before it is presented to the agent for scoring and drafting.

social