making-of
Pass
Audited by Gen Agent Trust Hub on Nov 11, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted chat transcripts to generate project code and execute rendering workflows. Malicious instructions embedded in the chat history could potentially influence the agent's behavior during these tasks. \n
- Ingestion points:
t3_thread_readand user-provided transcripts. \n - Boundary markers: None. \n
- Capability inventory:
npm install,npx remotion,ffmpeg,sips, andtailscale. \n - Sanitization: None. \n- [COMMAND_EXECUTION]: The workflow involves executing numerous command-line utilities including
npm,npx,ffmpeg,sips, andtailscaleto process and serve media based on external input. \n- [EXTERNAL_DOWNLOADS]: The project scaffolding process usesnpm installto download dependencies from the npm registry. \n- [DYNAMIC_EXECUTION]: The skill generates a TypeScript file (src/story.ts) based on chat content, which is subsequently bundled and executed by the Remotion rendering engine.
Audit Metadata