The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides example shell commands to fetch news items and social media content from well-known public services, including Google News, Algolia (Hacker News API), and Reddit. These commands use standard utilities like curl and pipe the resulting data to parsers such as jq and xmllint for structural processing.
[EXTERNAL_DOWNLOADS]: The skill references and integrates data from well-known external technology services and APIs to support media monitoring and industry research. It also includes a citation link to an external repository on GitHub (github.com/Knowatoa/ai-visibility-skills) for source attribution.
[INDIRECT_PROMPT_INJECTION]: The skill interacts with untrusted external data as a primary function of its newsjacking and podcast preparation workflows.
Ingestion points: Data is fetched from external RSS feeds and news headlines via shell commands defined in references/newsjacking.md and through browser-based research described in references/podcast-guest-prep.md.
Boundary markers: The instructions do not define specific delimiters or isolation techniques for the ingested text within prompts, relying on user review of the drafted pitches.
Capability inventory: The agent has capabilities for network access (via curl) and local file storage within the .agents/ directory for tracking media lists and marketing context.
Sanitization: The skill uses structural parsers (jq, xmllint) to handle raw data formats, though the natural language content extracted from these sources is not subject to further sanitization before being processed by the agent.

public-relations