remotion-graphics
Fail
Audited by Gen Agent Trust Hub on Nov 9, 2026
Risk Level: HIGHEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [REMOTE_CODE_EXECUTION]: The skill instructs the agent to clone a third-party GitHub repository (
https://github.com/agamm/video-agent) and run the agent within that environment. This involves executing scripts and code from an external source that is not verified as a trusted vendor or well-known service. - [EXTERNAL_DOWNLOADS]: Executing the Remotion render commands triggers an automatic download of a headless Chromium browser binary (~150–300 MB) from external servers. The skill also requires running
npm install, which downloads numerous packages from the public npm registry. - [COMMAND_EXECUTION]: The skill's workflow depends on executing multiple shell commands, including
npm install,npx remotion render, andffmpeg. These commands are used to manage dependencies and perform heavy video processing tasks. - [INDIRECT_PROMPT_INJECTION]: The skill processes external transcript data and cue sheets to determine the timing and content of animated graphics, creating a potential surface for indirect prompt injection.
- Ingestion points: Transcription data (
--wordstranscript) and cue sheets parsed byFinalEdit.tsxin the referenced repository. - Boundary markers: No explicit delimiters or instructions are provided to the agent to treat transcript content as untrusted data.
- Capability inventory: The agent has access to shell execution via
npm,npx, andffmpegto process the data. - Sanitization: There is no evidence of sanitization or validation of the transcript content before it is interpolated into animation frame calculations and command arguments.
Recommendations
- AI detected serious security threats
Audit Metadata