tesseract-video
Pass
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPRIVILEGE_ESCALATIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill downloads the Tesseract CLI binary and its corresponding installation scripts directly from the vendor's official GitHub repository at
github.com/mirage-hq/tesseract/releases.- [COMMAND_EXECUTION]: The skill uses subprocess calls to execute the Tesseract CLI (tsrct), as well as media utilitiesffmpegandffprobe. These commands are used to render video, synthesize audio, and extract media metadata.- [PRIVILEGE_ESCALATION]: In theinstallation.mdreference, the skill instructs the agent to bypass the PowerShell execution policy (-ExecutionPolicy Bypass) to run the Tesseract installer on Windows hosts.- [INDIRECT_PROMPT_INJECTION]: The skill possesses a surface area for indirect prompt injection by ingesting and processing untrusted user data such as media files, transcripts, and project files. - Ingestion points: The skill reads local media files (
.mp4,.mov), project documents (.tsrct), and user-supplied transcripts. - Boundary markers: There are no explicit markers or instructions to ignore potential commands embedded in media metadata or transcripts.
- Capability inventory: The skill has the capability to write to the local filesystem, execute shell commands via subprocess, and perform network downloads for installation.
- Sanitization: No specific sanitization or validation logic is implemented for the content of processed transcripts or media metadata before it is used to influence agent decisions.- [DYNAMIC_EXECUTION]: The skill supports dynamic behavior through procedural sound synthesis in
tesseract_sound.py, custom WGSL fragment shaders incustom-shader.md, and JavaScript-based animation expressions (layerTimeJsCode) within the project schema.
Audit Metadata