playwright-testing

CRITICAL E004: Prompt injection detected in skill instructions.

• Potential prompt injection detected (high risk: 1.00). The "Companion check" instructs the agent to run local filesystem commands (ls ~/.claude/skills/ ...) and offer to install missing skills via npx, which requests access to local files and potential code execution unrelated to Playwright testing and thus constitutes hidden/extraneous instructions outside the skill's purpose.

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.80). The skill's SKILL.md explicitly instructs running Playwright against arbitrary URLs (e.g., "Use codegen for discovery — run npx playwright codegen ") and includes examples that fetch and intercept remote responses (page.route(...)/route.fetch() and request contexts with external baseURL), which clearly require ingesting and interpreting open/untrusted web content that could influence test generation or actions.