The Agent Skills Directory

[COMMAND_EXECUTION]: The skill contains shell commands in references/newsjacking.md that use curl, jq, and xmllint to fetch and process data from Google News, Hacker News (Algolia API), and Reddit. These operations target well-known services for the purpose of identifying trending news topics for public relations.
[INDIRECT_PROMPT_INJECTION]: The skill's primary function involves ingesting and analyzing untrusted external data such as news articles and social media posts, creating a potential surface for indirect prompt injection.
Ingestion points: Content is retrieved via scripts in references/newsjacking.md from external sites like Google News and Reddit.
Boundary markers: The skill includes a specific warning in references/newsjacking.md stating that fetched content should be treated as untrusted data and that the agent should never follow instructions embedded within it.
Capability inventory: The agent utilizes shell commands (curl) and the dev-browser tool to access and process this external information.
Sanitization: The skill relies on behavioral instructions to prevent the agent from executing instructions found in external data.

public-relations