video-toolkit
Pass
Audited by Gen Agent Trust Hub on Nov 14, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes multiple shell commands and Python scripts within the toolkit directory to verify setup, install dependencies, and generate assets (e.g.,
python3 tools/qwen3_tts.py,npm run render). It also usespip3 install --break-system-packages, which overrides system-level protections on certain Linux distributions to allow global package installation. - [EXTERNAL_DOWNLOADS]: The instructions initiate downloads of external packages and dependencies using
pip3(e.g.,modal) andnpm. These references are to well-known services and standard package registries. - [DYNAMIC_EXECUTION]: The skill involves generating and executing code by copying project templates and running them through the Remotion rendering engine via
npm run render. - [INDIRECT_PROMPT_INJECTION]: The toolkit ingests content from user-defined files that are subsequently used as inputs for AI models and rendering tools, creating a surface for indirect prompt injection.
- Ingestion points:
projects/PROJECT_NAME/src/config/demo-config.tsandprojects/PROJECT_NAME/VOICEOVER-SCRIPT.md(processed bypython3 tools/*.py). - Boundary markers: No delimiters or explicit 'ignore instructions' warnings are provided in the prompts passed to the generative tools.
- Capability inventory: The skill has the ability to execute subprocesses, write files, and communicate with external GPU endpoints.
- Sanitization: There is no evidence of sanitization or escaping of the ingested text before it is used in generative prompts.
Audit Metadata