The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted data from external social media platforms such as Reddit, X, LinkedIn, and Hacker News for social listening purposes. This constitutes an indirect prompt injection surface where malicious instructions could be embedded in the fetched content. 1. Ingestion points: Data is retrieved via shell commands (curl) and browser automation (dev-browser) from various public APIs and web pages as described in 'references/listening.md'. 2. Boundary markers: The skill includes a clear safety warning in 'references/listening.md' stating that fetched content is untrusted data and the agent must never follow instructions embedded in post text or comments. 3. Capability inventory: The skill has access to shell execution for data retrieval and browser navigation tools, and references potential integration with social media scheduling APIs. 4. Sanitization: The instructions provide a scoring rubric and triage process that directs the agent to analyze the content's sentiment and fit rather than executing it as instructions.
[COMMAND_EXECUTION]: The skill provides specific shell command templates in 'references/listening.md' using curl, jq, and xmllint to fetch and parse JSON or RSS data from well-known services (Reddit, Algolia, Bluesky). These are legitimate data retrieval operations required for the skill's primary functionality.

social