filmcraft-cli-media
Pass
Audited by Gen Agent Trust Hub on Oct 12, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill downloads the FilmCraft CLI binary and desktop application (DMG) from trusted GitHub repositories associated with the vendor and verified organizations. These downloads are strictly validated against hardcoded SHA-256 checksums and specific URL path patterns to prevent tampering or redirected downloads.
- [COMMAND_EXECUTION]: Native binaries are executed via the Python subprocess module. To mitigate risk, the skill wraps these executions in a macOS sandbox (
sandbox-exec), which restricts the process's file system access and network capabilities to only those explicitly authorized by the user through command-line arguments. - [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied media files and project configuration data which could theoretically contain malicious instructions.
- Ingestion points: Processes JSON project plans and media files (MP4, WAV, PNG, etc.) from paths specified by the user in the
--read-rootparameter. - Boundary markers: Implements structured JSON schemas (
craft-command-plan/v1) and performs strict validation on sensitive parameters, such as ensuring timing values are exact 64-bit integers. - Capability inventory: The skill can perform file system operations (read/write), network downloads (limited to trusted sources), and execute subprocesses within a restricted environment.
- Sanitization: Uses SHA-256 hash verification for all input assets and enforces an operating system-level sandbox to isolate the media processing environment from the rest of the system.
Audit Metadata