The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill fetches public data from well-known services and platforms, including Reddit, Hacker News, Bluesky, and YouTube, to facilitate its social listening and engagement triage features.
[COMMAND_EXECUTION]: Provides pre-defined shell recipes in references/listening.md that use curl, jq, and xmllint to retrieve and parse data from external web APIs.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted content from social media platforms to perform scoring and draft replies, which creates a vulnerability to indirect prompt injection.
Ingestion points: Untrusted data enters the agent context via the curl commands defined in references/listening.md which pull content from Reddit, Hacker News, and Bluesky.
Boundary markers: The skill does not define specific delimiters or "ignore" instructions to separate external content from the agent's system instructions during the daily triage loop.
Capability inventory: Across its scripts, the skill utilizes network operations (curl) and reads local project configuration files (e.g., product-marketing.md and listening-sources.md).
Sanitization: No explicit sanitization, filtering, or escaping of the external content is performed before the agent processes it for scoring and drafting.

social