ClawSight logoClawSight

ClawSight

An open-source Endpoint Detection and Response platform for AI Agent Security.

Real-time visibility, insight and control.

terminal
$ npx -y @cantinasecurity/clawsight install --mode local --link

Runs in seconds with OpenClaw builds

Crab claw top pincer
Crab claw bottom pincer

Product

Explore the ClawSight Experience

A unified dashboard for real-time security intelligence across your entire AI stack.

Dashboard

Last 24h · 3 agents

Total Activity

Telemetry events per hour for the last 24 hours

0408012016021:0001:0005:0009:0013:0017:00

Events / Blocks / Alerts

Total distribution in the last 24 hours

Events518
Blocks4
Alerts5
Total: 527

Top Called URLs

URL to resolved IP mapping (24h)

https://www.theguardian.com/world

151.101.133.111

6

https://apnews.com/

104.16.22.8

6

https://stooq.com/q/d/l/?s=cl.f&i=d

78.47.75.66

6

https://www.bbc.com/news

199.232.32.81

6

https://www.reuters.com/world/

108.157.128.48

6

Top Telemetry Categories

Most frequent categories in the last 24 hours

session
199
tool
132
policy
67
diagnostic
59
agent
36
message
17
gateway
8

OpenClaw telemetry operations

Use Cases

Tailored to Your Workflow

Whether you're an organization running agents at scale or an individual shipping with AI, ClawSight gives you real-time visibility, personalized rules, and threat detection right where you work.

Track and quantify agent behaviors in production

Monitor every action your AI agents take, from tool calls to output generation. Get real-time alerts when behavior deviates from expected patterns.

  • Detect unsafe actions before they cause damage
  • Trace decision paths across multi-step workflows
  • Monitor output drift and hallucination patterns
AI Agent Monitoring
>agent: gpt-4o-sec | status: active
>actions: 1,247 | flagged: 3
>drift score: 0.02 (stable)
Streaming live data

Features

Built Different for AI Security

Purpose-built capabilities that go beyond traditional SIEM to protect modern AI infrastructure.

AI-First Signal Detection

Sensors tuned for AI patterns traditional SIEMs miss. Detect prompt injection, model drift, and anomalous agent behavior out of the box.

Full Stack Observability

Unified view across prompts, API calls, agents, and infrastructure. One dashboard for your entire AI security posture.

Real-Time Risk Scoring

Dynamic threat levels and predictive flagging that adapt to your environment. Know your risk posture at every moment.

Zero-Config Deployment

Launch within minutes using OpenClaw. No complex setup, no lengthy onboarding — just instant security visibility.

Cantina logoCantina

Built by the Experts at Cantina

ClawSight is crafted by Cantina - the leading platform in security, combining advanced AI and domain expertise to find real risk, reduce noise, and accelerate delivery.

Enterprise

Need an In-Depth AI Security Scan?

If you're an enterprise and want a deep AI security assessment across your codebase, agents, and infrastructure, talk to our team today.