video-editing
Pass
Audited by Gen Agent Trust Hub on Jul 29, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [COMMAND_EXECUTION]: The skill includes numerous examples of FFmpeg commands and shell scripts for deterministic video editing tasks such as trimming, concatenating, and reframing footage.
- [EXTERNAL_DOWNLOADS]: The workflow integrates with well-known services including ElevenLabs for text-to-speech generation and fal.ai for creating media assets. It also utilizes
npxto execute the Remotion rendering engine. - [CREDENTIALS_UNSAFE]: Credential management follows best practices by demonstrating the use of environment variables (
os.environ) in Python scripts to handle API keys securely. - [INDIRECT_PROMPT_INJECTION]: The skill describes using LLMs to analyze video transcripts for identifying segments and generating edit decision lists.
- Ingestion points: User-provided video transcripts processed in Layer 2 (SKILL.md).
- Boundary markers: None are present in the example prompt templates.
- Capability inventory: File system access via FFmpeg, local script execution via Remotion, and network communication via API calls.
- Sanitization: No explicit content filtering or validation is specified for the input transcripts.
Audit Metadata