lighthouse-a11y

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

• Third-party content exposure detected (medium risk: 0.30). The skill’s runtime workflow (via npx lhci autorun) collects and audits the configured ci.collect.url pages with Lighthouse (HTML/content rendered from those URLs), so outsider-authored text embedded in those pages can be ingested and evaluated by the LLM indirectly through page content.

MEDIUM W013: Attempt to modify system services in skill instructions.

• Attempt to modify system services in skill instructions detected (high risk: 0.70). The skill's example configuration sets chromeFlags: '--no-sandbox', which explicitly bypasses Chrome's sandboxing (a security mechanism) and could compromise the host running the agent.