threejs-qa-release
Pass
Audited by Gen Agent Trust Hub on Jul 9, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [COMMAND_EXECUTION]: The skill executes a bundled Node.js script (
scripts/inspect-threejs-canvas.mjs) to automate browser interactions and verify game state. This is a legitimate part of the QA workflow described in the skill. - [EXTERNAL_DOWNLOADS]: The browser automation script depends on the Playwright library. When installed or executed, this library may download browser binaries (such as Chromium) to facilitate testing. These are downloads from a well-known, trusted source (Microsoft).
- [PROMPT_INJECTION]: The skill has an indirect prompt injection surface because it reads data from the web page being tested (specifically the
window.__THREE_GAME_DIAGNOSTICS__object). - Ingestion points: The
inspect-threejs-canvas.mjsscript retrieves data from the browser context and returns it to the agent. - Boundary markers: None identified for the browser-retrieved data.
- Capability inventory: The agent can execute local build commands and shell scripts.
- Sanitization: The data from the browser is captured and reported as JSON without explicit sanitization, though it is intended for diagnostics.
- [SAFE]: The skill includes security best practices in its release checklist, specifically instructing the agent to verify that no API keys or debug tools are accidentally exposed in the production build.
Audit Metadata