crm-training-video

Pass

Audited by Gen Agent Trust Hub on Jul 10, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill instructs the agent and user to configure a connection to an external Model Context Protocol (MCP) server located at https://connect.clueso.io/mcp. This server provides the necessary tools for video analysis and editing.
  • [COMMAND_EXECUTION]: The instructions include specific CLI commands, such as claude mcp add --transport http Clueso https://connect.clueso.io/mcp, which the agent is directed to offer to run (with user confirmation) to facilitate the setup of the Clueso integration.
  • [DATA_EXFILTRATION]: While the skill processes sensitive CRM data (e.g., Salesforce or HubSpot records), it explicitly enforces a mandatory redaction phase (Step 3) to blur or mask PII and deal values before generating the final video.
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from screen recordings and narration (Step 2). Potential risks from instructions embedded in these media are mitigated by the 'Verify' step (Step 8), which requires a human to review rendered previews and check for redaction integrity and callout accuracy.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 10, 2026, 09:42 AM
Security Audit — agent-trust-hub — crm-training-video