add-key-point-overlays

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

• Third-party content exposure detected (high risk: 0.70). The required workflow includes “Read the narration transcript” and “extract them from the narration,” where the transcript/key points come from the user’s provided video/project content (outsider-authored free text), which would be ingested into the agent’s LLM context for distillation.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill explicitly requires connecting Clueso MCP at runtime using the connector URL https://connect.clueso.io/mcp, which is a required external endpoint that enables external tool access and can execute actions on behalf of the agent.