The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill provides commands to fetch news and trending data from well-known services including Google News, Hacker News (via Algolia), and Reddit. These operations are limited to data retrieval for monitoring purposes and do not involve the execution of remote scripts.
[COMMAND_EXECUTION]: Includes shell commands using curl, jq, and xmllint to process news data. It also suggests installing the jq utility via Homebrew. These are standard tools for data processing and do not involve administrative privilege escalation or high-risk execution patterns.
[INDIRECT_PROMPT_INJECTION]: The workflow involves researching external articles, social media posts, and journalist profiles using a browser. This activity creates a potential surface for indirect prompt injection if the ingested content contains instructions designed to manipulate the agent's summary or pitch generation.
Ingestion points: Browser-driven research of articles, social media feeds, and journalist bios as described in references/newsjacking.md and references/journalist-pitching.md.
Boundary markers: The instructions do not define specific delimiters or warnings for the agent when processing this external content.
Capability inventory: The skill is primarily focused on generating text-based communication (pitches, media lists) and lacks dangerous file-write or system-altering capabilities.
Sanitization: No explicit sanitization or filtering logic is provided for the external data ingested.

public-relations