chrome-extension-dev

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 0.90). The documentation instructs adding an always-on "dev bridge" that forwards arbitrary page postMessage payloads to the extension and relays responses without origin/auth checks—this is a high-risk backdoor/data-exfiltration vector if left in non-dev builds.

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

• Third-party content exposure detected (medium risk: 0.30). The required runtime workflow instructs the agent to navigate to user-selected web pages (step “Navigate to test page”) and then read/inspect them via MCP tools like mcp__claude-in-chrome__read_page/console at runtime, meaning outsider-authored free text from those pages can be ingested.