The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from external social media platforms (Reddit, Hacker News, Bluesky, LinkedIn, and Twitter/X), which creates a surface for indirect prompt injection where malicious instructions in a post could influence the agent's behavior during triage or drafting.
Ingestion points: Data is ingested from public APIs and RSS feeds using curl and from authenticated social media feeds via the dev-browser tool, as detailed in references/listening.md and references/listening-sources-template.md.
Boundary markers: The skill includes logic for filtering and scoring posts, but does not implement explicit prompt boundaries or instructions to ignore commands embedded within the fetched social media text.
Capability inventory: The skill utilizes curl for fetching remote data and dev-browser for navigating and reading content from web platforms.
Sanitization: No sanitization, escaping, or validation of the external social media content is specified before the data is processed by the agent to draft responses or score engagement opportunities.

social