playwright-triage
Pass
Audited by Gen Agent Trust Hub on Nov 2, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- Indirect Prompt Injection: The skill is designed to triage and fix external GitHub issues, which serves as an ingestion point for untrusted data. While the instructions include guardrails, such as refusing to run PRs that modify sensitive configuration files, processing arbitrary issue descriptions is a potential surface for instructions that could influence agent behavior.
- Dynamic Execution: The skill involves generating and executing reproduction scripts within a scratch directory. This is used to verify fixes by running the MCP server CLI and calling tools via the SDK in a local environment.
- External Downloads and Command Execution: The skill utilizes standard development tools like
git,npm, andnpxto manage repositories and dependencies. It fetches source code from the official repository and installs Playwright-related packages. These operations target resources belonging to the skill's author and other well-known services.
Audit Metadata