Skills
skills/gitroomhq/agent-media/Make Product In Hands/Gen Agent Trust Hub

Make Product In Hands

Pass

Audited by Gen Agent Trust Hub on Jun 1, 2026

Risk Level: SAFE
Full Analysis
  • [PROMPT_INJECTION]: The skill processes untrusted input through character_sheet_url, product_image_url, and product_image_base64 parameters. This creates an indirect prompt injection surface where instructions could theoretically be embedded in the media metadata or contents.
  • Ingestion points: SKILL.md (as parameters for the mcp__agent-media__make_product_in_hands tool).
  • Boundary markers: Not explicitly defined in the skill instructions.
  • Capability inventory: The skill invokes the mcp__agent-media__make_product_in_hands tool and performs network requests to api.agent-media.ai.
  • Sanitization: The skill description notes that media is automatically re-hosted to R2, which may involve server-side processing and validation.
  • [DATA_EXFILTRATION]: The skill communicates with api.agent-media.ai to submit video generation requests and poll for results. This domain is the official service endpoint for the agent-media platform used by the skill.
  • [CREDENTIALS_UNSAFE]: The skill uses a placeholder variable $AGENT_MEDIA_API_KEY for authentication, following secure practices for secret management by not hardcoding sensitive tokens.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 1, 2026, 11:58 AM
Security Audit — agent-trust-hub — Make Product In Hands