The Agent Skills Directory

[PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface because it is designed to ingest and process data from external, untrusted sources such as RSS feeds and social media APIs.
Ingestion points: The skill fetches news data from Google News (RSS), Hacker News (via Algolia's API), and Reddit as documented in references/newsjacking.md.
Boundary markers: There are no explicit instructions or delimiters used to warn the agent to ignore potentially malicious instructions embedded in the news titles or article content it retrieves.
Capability inventory: The agent has the ability to read project-specific marketing context files (.agents/product-marketing.md), perform browser-based research on journalists, and generate outreach content (emails/pitches) for the user. It also maintains local media lists in the .agents/ directory.
Sanitization: No specific sanitization or filtering logic is provided for the data retrieved via the documented shell commands.
[COMMAND_EXECUTION]: The skill provides several shell command templates in references/newsjacking.md for news monitoring. These commands use curl to retrieve data from well-known services (Google, Reddit, and Algolia/Hacker News) and pipe the output to parsing tools like jq and xmllint. These are standard, benign data-gathering operations intended to facilitate the skill's primary purpose of media monitoring.

public-relations