The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill implements a social listening workflow that ingests content from external platforms such as Reddit, Hacker News, and Bluesky (references/listening.md). This constitutes an attack surface where malicious instructions embedded in social media posts could be processed by the agent.
Ingestion points: External data is fetched via curl from public JSON and RSS endpoints.
Boundary markers: The instructions do not specify explicit delimiters or "ignore instructions" warnings for the fetched content.
Capability inventory: The skill utilizes curl for network access and jq/xmllint for data parsing, with the agent instructed to analyze and draft responses to this content.
Sanitization: There is no explicit mention of sanitizing or filtering the external content before the agent processes it.
[EXTERNAL_DOWNLOADS]: The skill utilizes curl recipes to fetch information from well-known technology services, including Reddit, Algolia (for Hacker News), and Bluesky (references/listening.md). These network operations are limited to official API endpoints for public data retrieval and do not involve sensitive local data or untrusted sources.

social