demo-cutdown
Pass
Audited by Gen Agent Trust Hub on Sep 26, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill transcribes and analyzes spoken audio from user-provided video files to map narrative beats and generate scripts.
- Ingestion points: Spoken audio transcription from source videos (SKILL.md, Workflow Step 2).
- Boundary markers: None present; the skill does not instruct the agent to ignore instructions embedded in the audio transcript.
- Capability inventory: File duplication and modification, script generation, audio generation, and video export (SKILL.md, Workflow Steps 5, 6, and 12).
- Sanitization: No sanitization or validation of the transcribed text is mentioned before it is used to decide the condensed narrative or write replacement scripts.
- [COMMAND_EXECUTION]: The skill provides specific CLI commands to install a Model Context Protocol (MCP) server.
- Evidence: Instruction to run
claude mcp add --transport http Clueso https://connect.clueso.io/mcpin Claude Code. - [EXTERNAL_DOWNLOADS]: The skill instructs users to install external plugins and connectors for various platforms.
- Evidence: References to ChatGPT plugins and Codex plugins (e.g.,
app-6a310833394081919705b0a3ee745267@openai-curated-remote) which are external to the skill package.
Audit Metadata