ai-video-to-talking
Pass
Audited by Gen Agent Trust Hub on Nov 5, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user data, including video, audio, and dialogue scripts, which establishes a surface for indirect prompt injection. \n
- Ingestion points: User-provided media files and dialogue text are ingested as core inputs for processing (described in SKILL.md and references/media-input.md). \n
- Boundary markers: The skill does not define specific delimiters or instructions to ignore embedded commands within user-provided content. \n
- Capability inventory: The skill uses shell tools like curl and ffprobe, accesses the file system, and communicates with the vendor API (referenced in references/api.md). \n
- Sanitization: No specific sanitization or validation logic is prescribed for handling user-provided strings or filenames before they are processed. \n- [COMMAND_EXECUTION]: The skill instructions provide shell command templates (e.g., for ffprobe and curl) that the agent is expected to populate with filenames or parameters. If user-controlled strings are interpolated into these commands without proper sanitization, it could lead to command injection. \n- [EXTERNAL_DOWNLOADS]: The skill facilitates the installation and use of official vendor resources, such as the syncsdk Python package and the @sync.so/mcp-server Node.js package, from standard registries to enable connection to the sync.so service.
Audit Metadata