The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructions direct the agent to use the newsjack CLI for campaign management (e.g., newsjack media-lists create, newsjack media-lists job). These are legitimate administrative and functional commands for the Medialyst service.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external sources, including web search results, news search results, and third-party journalist profiles, creating a potential surface for indirect prompt injection.
Ingestion points: Data enters the agent's context through news-search tools, web search results, and the Medialyst API results (journalist bylines, roles, and recent work).
Boundary markers: The instructions do not specify technical delimiters (like XML tags) for external data, but provide conceptual boundaries like "Anti-hallucination" and "Human-send" rules.
Capability inventory: The skill uses CLI execution (newsjack) and network-enabled search tools to fulfill its purpose.
Sanitization: The skill mandates human review for all outreach ("Human-send") and requires the agent to verify identity, contacts, and bylines rather than guessing ("Anti-hallucination"), which significantly mitigates the impact of potential injection.

find-journalists