Gk0Wk

Gk0Wk / ClawGuard

Public

The antivirus for OpenClaw — approve dangerous actions, scan skills, block secret leaks, and keep humans in control, for safety.

clawguard.top agent-security ai-security audit-logs cybersecurity data-loss-prevention
25
0
69% credibility
Found Mar 13, 2026 at 25 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
TypeScript
AI Summary

ClawGuard is a security layer for OpenClaw that detects and pauses risky actions like dangerous executions, outbound messages, and file changes for human approval and auditing.

How It Works

1
🔍 Discover ClawGuard

You hear about ClawGuard, the safety shield that protects your AI assistant from risky moves like dangerous commands or secret leaks.

2
🛡️ Add the safety guard

You easily add ClawGuard to your AI assistant to watch over actions and keep things secure.

3
🔄 Refresh your assistant

You refresh your AI setup, and new safety pages appear for settings, approvals, and reviews.

4
⚠️ Spot a risky action

When you try something potentially harmful, like a big delete or sending sensitive info, ClawGuard pauses it and asks for your okay.

5
Review the alert
👍
Approve it

You give the okay once, and it proceeds safely after a quick retry.

👎
Block it

You deny it, and it stays stopped with a full record.

6
📊 Check the records

You visit the audit page anytime to see a clear history of decisions and what happened.

🛡️ Stay in control

Now your AI assistant runs safely, with you always having the final say on risks.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 25 to 25 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is ClawGuard?

ClawGuard is a TypeScript plugin for OpenClaw that acts as an antivirus layer, intercepting risky actions like shell exec, outbound messages, and workspace writes/edits. It scans tool params for dangerous commands, secret leaks, suspicious paths, and public destinations, then queues them for human approval via dedicated pages at /plugins/clawguard/approvals, /audit, and /settings. Install it locally with `openclaw plugins install ./plugins/openclaw-clawguard` for immediate safety controls and audit trails—far from clawguard tape or door shields, this guards AI agents.

Why is it gaining traction?

It hooks directly into OpenClaw's before/after-tool and message_sending events, offering fast-path blocks or one-time approvals without slowing safe flows, unlike generic github antivirus apk or tron antivirus github tools. Developers notice instant protection for linux exec risks, windows commands, or secret scans in outbound actions, with clear reasons like "recursive delete detected." The plugin-hosted UI and persisted state make reviews practical, standing out in antivirus github android/linux spaces.

Who should use this?

OpenClaw users running community skills or group-chat agents, especially ops teams handling untrusted inputs that trigger exec or sends. Ideal for devs evaluating github antivirus remover/scan needs in agent workflows, or anyone blocking secret leaks in workspace mutations. Skip if you're not on OpenClaw.

Verdict

Worth a local demo install for OpenClaw safety (17 stars, 0.7% credibility score), with strong docs-first approach and working approvals/audit despite sprint-0 limits like minimal outbound coverage. Test it before production—mature enough for risk-aware pilots, not broad GA yet. (198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.