ugc-testimonial-video
Pass
Audited by Gen Agent Trust Hub on Sep 8, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill provides a setup command to add a Model Context Protocol (MCP) server from the vendor's official domain (
https://mcp.apimage.org/mcp). This is used to extend the agent's capabilities with video generation tools. - [INDIRECT_PROMPT_INJECTION]: The skill's primary function is to process external data (customer reviews, emails, and support tickets) into video scripts. This creates a surface for indirect prompt injection if the ingested data contains instructions designed to manipulate the LLM.
- Ingestion points: Customer reviews collected from emails, support tickets, and marketplace reviews as described in the workflow section of
SKILL.md. - Boundary markers: None explicitly defined in the provided workflow to delimit the untrusted review text.
- Capability inventory: The skill utilizes a
generate_videotool (configured via MCP) to produce media based on the processed inputs. - Sanitization: No specific sanitization or filtering logic is described for the incoming review text beyond the instruction to keep quotes verbatim.
Audit Metadata