# ClawGuard

- **Event:** [OpenClaw Hackathon](https://cerebralvalley.ai/e/open-claw-hackathon-41d0b52d)
- **When:** Sun, Feb 15 at 1:00 – 6:00 PM (PST)
- **Where:** Frontier Tower 🧑‍🚀, San Francisco, CA
- **Team:** [Max Xie](https://cerebralvalley.ai/u/maxxie114), [Kush Ise](https://cerebralvalley.ai/u/KUSH2704)
- **GitHub:** https://github.com/maxxie114/ClawGuard
- **Website:** http://157.230.149.230:8000/
- **Demo video:** https://youtu.be/Xzg6fOQOluU
- **Gallery:** https://cerebralvalley.ai/e/open-claw-hackathon-41d0b52d/hackathon/gallery
- **Page:** https://cerebralvalley.ai/e/open-claw-hackathon-41d0b52d/hackathon/gallery/8

ClawGuard is a security-first filtering server that sits between external inputs (email, webhooks) and LLM agents, preventing prompt injection, secret exfiltration, and malicious content from ever reaching the AI. It connects to Gmail via GCP Pub/Sub for real-time email ingestion, runs a deterministic sanitization pipeline that strips HTML/scripts, detects 10+ injection patterns, redacts secrets, removes hidden unicode characters, and scores every
  message with a 0-100 risk rating. Built for OpenClaw, ClawGuard exposes a REST API for agent skills to query sanitized  emails on demand, and includes a live dashboard that visually demonstrates the firewall in action — showing raw vs.  sanitized content side-by-side with color-coded risk highlighting.

---

Markdown version of https://cerebralvalley.ai/e/open-claw-hackathon-41d0b52d/hackathon/gallery/8. Site index for agents: https://cerebralvalley.ai/llms.txt · full text: https://cerebralvalley.ai/llms-full.txt