carocut-reviewer

Pass

Audited by Gen Agent Trust Hub on Jun 16, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill utilizes several standard development tools via shell commands, including npx tsc for TypeScript verification, npx remotion for video rendering, and ffprobe for verifying media output metadata.
  • [COMMAND_EXECUTION]: Instructions include the execution of a repository-local script sh scripts/browser_install.sh used for environment setup.
  • [EXTERNAL_DOWNLOADS]: The skill references npx remotion browser-ensure, which automates the download of Chromium binaries required by the Remotion engine for rendering. This is a well-known and expected behavior for this technology stack.
  • [PROMPT_INJECTION]: A surface for indirect prompt injection exists as the agent is instructed to process user feedback and analyze project source files (.tsx, .ts). This is a legitimate functional requirement for a 'reviewer' skill, and the skill suggests structured communication formats (JSON) which help mitigate accidental instruction following from untrusted data.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 16, 2026, 09:05 AM
Security Audit — agent-trust-hub — carocut-reviewer