resemble-detect

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.90). The skill explicitly requires ingesting media hosted at publicly accessible HTTPS URLs (see "Required Setup: Media Requirement" and the "Submit a Detection" examples) and can enable reverse image search / external source inspection (see use_reverse_search and reverse_image_search_sources), so the agent will read and act on untrusted third‑party content that can influence decisions.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.80). The skill explicitly directs agents to connect at runtime to the hosted MCP SSE endpoint (https://mcp.resemble.ai/sse), which supplies live tools and documentation (e.g., resemble_docs_lookup, resemble_api_endpoint) that the agent will use to shape prompts and behavior, so this external URL can directly control agent instructions.