experience-content-media-search
Pass
Audited by Gen Agent Trust Hub on Jun 26, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill enforces a strict multi-step human-in-the-loop (HITL) process. It explicitly prohibits the agent from calling search tools autonomously, requiring it to first present available sources to the user and wait for a selection.
- [SAFE]: It utilizes a pattern of tool introspection where the agent checks its context for specific MCP tools (
search_media_cms_channelsandsearch_electronic_media) before presenting options to the user, ensuring that only accessible and authorized search methods are offered. - [SAFE]: Data handling is constrained to constructing structured JSON search payloads for vendor-specific tools. The skill provides detailed templates for keyword and taxonomy extraction, which reduces the risk of malformed inputs while maintaining the functionality of the media search service.
Audit Metadata