The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze data from untrusted external sources, which could potentially contain malicious instructions intended to manipulate the agent's behavior.
Ingestion points: The skill fetches data from Google News RSS feeds, the Hacker News API, and Reddit (via references/newsjacking.md), as well as podcast RSS feeds (via references/podcast-guest-prep.md).
Boundary markers: The instructions lack explicit delimiters or "ignore previous instructions" warnings when processing the interpolated external data.
Capability inventory: The skill uses curl for network requests and dev-browser for web research, which are used to retrieve the external content.
Sanitization: There is no evidence of content sanitization or validation before the external data is processed by the agent.
[UNVERIFIABLE_DEPENDENCIES_AND_REMOTE_CODE_EXECUTION]: The skill utilizes external utilities and fetches data from remote servers.
Package Installation: The file references/newsjacking.md recommends the installation of the jq utility using the command brew install jq.
Remote Data Access: The skill provides examples of retrieving news data from well-known platforms including Google News (news.google.com), Algolia's Hacker News API (hn.algolia.com), and Reddit (www.reddit.com).

public-relations