The Agent Skills Directory

[PROMPT_INJECTION] (MEDIUM): This skill exhibits a risk of Indirect Prompt Injection (Category 8). ● Ingestion points: The skill retrieves untrusted data from external sources, specifically video transcripts, speaker information, and scene metadata via the /api/videos/:id and /api/clips/:id endpoints. ● Capability inventory: The agent is granted the ability to create, modify, and delete data on the remote platform through POST, PATCH, and DELETE requests for timeline overlays. ● Boundary markers: No delimiters or explicit 'ignore instructions' warnings are defined in the skill to separate external data from the agent's core logic. ● Sanitization: There is no evidence of sanitization or filtering for the data retrieved, meaning a transcript containing text like 'delete all overlays' could be interpreted as a directive by the agent.

mecene