Security Layer for OpenClaw web agents
Know when your OpenClaw agents go rogue. Detect anomalies, track agent behavior, and get notified before critical events become incidents.
Powered by
Visual agent trace graph
See exactly how your OpenClaw agents connect, communicate, and execute — traced in a real-time node graph.
> main
> 05:38:01 PM
Hey! Welcome back — we're still on square one here, which is actually kind of exciting. Fresh start....
> 05:48:38 PM
HEARTBEAT_OK
> 06:48:38 PM
1.8s
HEARTBEAT_OK
Everything you need to monitor AI agents
Built for teams running production OpenClaw agents. From real-time observability to automated incident response.
Full Event Visibility
Every query, API call, file access, and shell execution is captured and indexed. Never miss what your agents are doing.
Intelligent Alerts
Custom rules engine that triggers alerts on suspicious patterns, rate limit breaches, PII exposure, and destructive commands.
Live Event Stream
Sub-second latency streaming of all gateway events. Watch agent behavior unfold in real-time from a single pane of glass.
Threat Detection
Automatic classification of malicious patterns including prompt injection attempts, data exfiltration, and privilege escalation.
Audit Trail
Immutable, tamper-proof logs for every agent action. Meet compliance requirements with queryable, exportable audit records.
Instant Response
Automatically block or quarantine agents when critical thresholds are breached. Configurable policies with kill-switch capability.
Catch threats before they become incidents
Configure granular alert rules for critical and malicious activity. Get notified via Slack, PagerDuty, email, or webhook in seconds.
Block and alert on rm -rf, DROP TABLE, and other destructive shell/SQL commands
Detect SSN, credit card, and email patterns in agent outbound responses
Alert when any agent exceeds 90% of configured token or request limits
Flag access attempts to restricted paths, APIs, or data sources
ML-based detection of unusual request patterns, timing, or payload sizes
Up and running in 4 simple steps
Get ClawTrace monitoring your OpenClaw agents in under a minute.
Open your OpenClaw terminal
Open the terminal on the machine where your OpenClaw agents are running, whether it's a Mac Mini, VPS instance, or Docker container.
Run the setup script for dashboard setup
One command auto-detects your OpenClaw gateway, pulls the config, and starts ClawTrace Dashboard. Just pass your OpenAI API key.
curl -fsSL https://raw.githubusercontent.com/dibbaa-code/clawtrace/main/setup.sh | bash -s -- OPENAI_API_KEY=<your-openai-api-key>Open in browser
Your dashboard is live. Access it at the URL shown in the terminal output.
http://<your-openclaw-machine-ip>:3000/monitorAdd alerting (Discord, WhatsApp, etc.)
When you're ready to get alerts, re-run the setup script with your webhook URL. ClawTrace will rebuild with alerting enabled.
curl -fsSL https://raw.githubusercontent.com/dibbaa-code/clawtrace/main/setup.sh | bash -s -- OPENAI_API_KEY=<your-key> DISCORD_WEBHOOK_URL=<your-webhook> CLAWTRACE_BASE_URL=http://<machine-ip>:3000