The Agent Skills Directory

[COMMAND_EXECUTION]: The files references/newsjacking.md and references/journalist-pitching.md contain shell command patterns, including instructions to install the jq utility via brew install jq and example curl commands with unvalidated placeholders like QUERY and SUBREDDIT.
[EXTERNAL_DOWNLOADS]: The skill retrieves data from external well-known services, including Google News (news.google.com), Hacker News (hn.algolia.com), and Reddit (reddit.com), for news monitoring purposes.
[PROMPT_INJECTION]: The skill's newsjacking workflow is susceptible to indirect prompt injection because it ingests and processes untrusted content from external articles and social media threads.
Ingestion points: External data is ingested through shell commands in references/newsjacking.md targeting various news RSS feeds and APIs.
Boundary markers: There are no explicit instructions or delimiters used to separate the ingested external text from the agent's core instructions, nor are there warnings to ignore embedded instructions.
Capability inventory: The agent has the capability to execute shell commands (curl, jq, xmllint), read local files, and draft outbound communication based on the ingested data.
Sanitization: The skill does not implement sanitization or validation of the content retrieved from external sources before it is analyzed to generate media pitches.

public-relations