playwright-debugger
Pass
Audited by Gen Agent Trust Hub on Nov 6, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes external CLI tools including
playwrightfor test reproduction andgh(GitHub CLI) for downloading artifacts. These executions are gated by safety procedures requiring user approval and repository trust. Therun-artifact-reader.shscript enforces an isolated environment and uses a root-owned system interpreter to run bundled Python scripts.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes end-to-end test artifacts which are considered untrusted data.\n - Ingestion points: Artifacts in
playwright-report/and CI run artifacts.\n - Boundary markers: Instructions to quote output and treat report-derived strings as untrusted.\n
- Capability inventory: Shell command execution, file system writes, and network operations via
gh.\n - Sanitization: Employs the
residual_credentials.pylibrary for recursive redaction of sensitive data.\n- [DYNAMIC_EXECUTION]: Thescripts/publish-json-report.pyscript usesimportlibto dynamically load theread-playwright-artifact.pyvalidator script. This is a local load of a bundled script within the skill's own directory, accompanied by path validation to ensure integrity.\n- [SAFE]: The skill implements robust security controls, including symlink rejection usingO_NOFOLLOW, path traversal prevention, and strict JSON parsing. Theresidual_credentials.pymodule provides a comprehensive redaction system for credentials and sensitive URL components.
Audit Metadata