We actively maintain the following branches:

Quill includes multiple layers of security by design:

• Sandboxed Execution — Code runs in isolated environments (local, Docker, or E2B sandboxes)
• Tool Annotations — Tools declare safety properties (read-only, destructive, idempotent, open-world)
• Adaptive Permissions — Progressive trust levels from Strict → Full based on session history
• Security Scanner — Deterministic offline scanning blocks malicious skills before any LLM call
• Depth-Aware Policy — Subagents lose dangerous tools as nesting depth increases
• Input Sanitization — Prompt injection defense at the middleware layer
• CSRF Protection — Double Submit Cookie pattern for all state-changing API calls

We take security vulnerabilities seriously. Please report them responsibly:

1. Do not open a public GitHub issue for security vulnerabilities.
2. Go to https://github.com/CreamyLong/quill/security/advisories to report privately.
3. Include a detailed description, reproduction steps, and potential impact.

We will acknowledge your report within 48 hours and provide a fix timeline within 7 days.

• Always use the latest release version
• Enable sandbox mode for code execution
• Use adaptive permissions to limit tool access
• Keep your API keys in .env (never commit them)
• Run behind a reverse proxy (Nginx) in production
• Enable authentication for multi-user deployments