recording-to-video-article
Pass
Audited by Gen Agent Trust Hub on Sep 26, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data by transcribing and analyzing user-provided screen recordings. This content is used to derive steps, draft documentation, and generate narration, creating a surface where malicious instructions embedded in audio or visual data could influence the agent's behavior.
- Ingestion points: Screen recordings (SKILL.md, Workflow Step 2).
- Boundary markers: None explicitly defined for the analysis of external recording data.
- Capability inventory: Project creation, article writing, and video rendering via Clueso MCP tool calls.
- Sanitization: No specific sanitization or filtering of transcribed content is mentioned.
- [EXTERNAL_DOWNLOADS]: The skill references an external configuration endpoint for the Clueso MCP connector at
https://connect.clueso.io/mcp. This resource belongs to the author's infrastructure and is required for the skill's primary functionality. - [COMMAND_EXECUTION]: The instructions suggest that the agent can execute a command to install the necessary MCP connector (
claude mcp add --transport http Clueso https://connect.clueso.io/mcp) upon receiving user confirmation.
Audit Metadata