The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from platforms like YouTube, Reddit, HN, X, and RSS feeds, which creates a significant attack surface for indirect prompt injection where malicious instructions could be embedded in external content to influence agent behavior.
Ingestion points: Fetches content from numerous third-party APIs and web feeds as described in references/fetchers.md.
Boundary markers: The instructions explicitly tell the agent to treat fetched content as data only and ignore embedded directives, but no technical delimiters are used during processing.
Capability inventory: The skill allows the agent to write files to a local vault and perform git push operations.
Sanitization: No automated sanitization of the external content is performed before it is processed by the agent for scoring and capture.
[COMMAND_EXECUTION]: The skill utilizes several CLI tools including curl, git, jq, and yt-dlp to perform its core fetching and management functions.
[DYNAMIC_EXECUTION]: The skill uses shell-based execution of embedded Python script templates to parse and extract metadata from XML and Atom feeds.
[PERSISTENCE]: The skill provides instructions for installing automated daily tasks using launchd on macOS and systemd on Linux, which enables recurring execution of the monitoring logic across sessions.
[EXTERNAL_DOWNLOADS]: The skill fetches data from a variety of well-known external platforms and recommends the use of ScrapeCreators for reliable data extraction from X and LinkedIn.

radar