ClawMetry for Managed OpenClaw
Works with every managed OpenClaw provider: KiloClaw, Hostinger, xCloud, MyClaw, Kimi Claw, and any self-hosted instance. One agent command is all you need.
ClawMetry for NemoClaw
Full observability for every NemoClaw sandbox. Live flow, token costs, brain tab, sub-agent tracking across your entire fleet. One dashboard for all nodes.
All channels supported
Telegram, iMessage, WhatsApp, Signal, Discord, Slack, BlueBubbles, Google Chat, MS Teams, Matrix, Zalo, WebChat. Live chat view with message counts for each. Auto-detected from your OpenClaw config.
Cloud version
See your agents from anywhere. Connect any OpenClaw instance with one command: clawmetry connect. E2E encrypted. Fleet dashboard for all your nodes.
Mac app
ClawMetry in your menu bar. Native Swift app. Monitor your agents without opening a browser. v1.0.1 available now.
Download for macOS →Every major agent runtime
OpenClaw, NemoClaw, Claude Code, Codex, Cursor, Hermes, Copilot, Grok, n8n, and more, all auto-detected into one dashboard. New runtimes ship every couple of weeks. See the full live list.
See every runtime →ClawMetry Device
Hardware companion for your fleet. Always-on glanceable status, physical kill switch, voice approvals from your desk. Ships with OTA self-update baked in.
See the device →Hosted evals + LLM judge
DeepEval-powered evals run in the cloud on every session. Multi-provider judge (Anthropic, OpenAI, Google) with an honest skip when a key is missing, so you never get a silent false pass.
Efficiency grade + waste meter
An A to F efficiency grade per runtime, with Cache-Hit and Routing-Advisor tiles. See exactly how much you'd save by tuning cache reuse, model choice, and prompt structure.
Windows & Linux desktop
Signed installers for every desktop OS: .dmg (macOS, notarized), .exe (Windows), .deb (Debian/Ubuntu). Thin-shell bundle auto-installs ClawMetry and self-updates every 6h.
OpenTelemetry receiver
Point any OTLP-instrumented agent at ClawMetry. The /v1/metrics, /v1/traces, and /v1/logs endpoints flow into the same DuckDB store as your OpenClaw sessions. Bring your own instrumentation, keep one dashboard.
EU AI Act compliance pack
Auto-generated audit trail, incident log, and Article 12/15/16 records for every agent action, exportable as a signed PDF for regulators. Docs & checklist live today; one-click export in progress.
Read the checklist →Security threat detection
Automated detection of compromised agents, prompt injection attacks, credential theft, and suspicious tool calls. BitSight found 30,000+ exposed OpenClaw instances — ClawMetry detects when something goes wrong.
Real-time cloud streaming
Watch your agent work live from anywhere in the world. The Flow tab animates in real-time on the cloud dashboard, same as local. WebSocket relay from your machine to any browser or native app.
iOS & Android
Check in on your agents from anywhere. Approve human-in-the-loop requests straight from your lock screen.
Alerting
Get paged when your agent is stuck, looping, or silent. PagerDuty, OpsGenie, Slack, Telegram, webhooks.
Human-in-the-loop
Pause an agent mid-task, see what it is about to do, then approve or deny from your dashboard or phone.
Framework SDK
Lightweight SDK for CrewAI, LangGraph, AutoGen and other non-OTel agent frameworks. (OTel-instrumented agents already flow in today. See the OpenTelemetry receiver card.)
Team features
Org shell & owner-based billing are LIVE (one payer, whole team observes). Shared dashboards, role-based access, and a full audit log of every agent action are next.
Cost analytics
Set budgets per agent, get alerted before you overspend, and see exactly where every token is going.
Token spike alerts
Set rules like "alert me if any session exceeds $5" or "notify when token velocity spikes 3x." Auto-detect runaway agents before the bill arrives.
Intent vs execution detection
Flag when an agent's stated reasoning doesn't match its actual tool calls. Catch the gap between what the agent said it would do and what it actually did.
Agent authority tracking
Track whether an agent was authorized to take the action it took. Integrates with OpenClaw exec approvals to show the gap between "did X" and "was allowed to do X."
External API tracing
Trace agent calls to external APIs via webhooks and HTTP callbacks. See the full chain when your agent talks to third-party services.
Cron calendar view
Visual calendar showing all scheduled cron jobs, their run history, and next execution times. See your agent's schedule at a glance.
Want to know when things ship?
Drop your email. We'll send one message when something on this list goes live. No marketing, no sequences.
Session replay
Step through any session chronologically with full cost attribution. See exactly what happened, in order, like replaying a video of your agent's work.
Live tool call stream
Real-time feed of every tool call across all active sessions. See exec commands, file reads, web fetches as they happen across your entire fleet.
Stuck session alerts
Detect when an agent is stuck, looping, or silent. Get notified before a runaway session burns through your API budget overnight.
30-day activity heatmap
Hourly usage grid showing when your agents are most active. Spot patterns, identify idle time, and optimize scheduling.
Smart model routing
Auto-downgrade cheap tasks to smaller models to save 10-50% on API costs. Rules-based routing that learns which tasks need GPT-4 vs which can use a cheaper model.
Emergency stop + spend alerts
Kill switch to stop all agent activity instantly. Set spend limits per agent, per day, per session. Auto-pause when budgets are hit.
Remote approval gates
Approve or deny irreversible agent actions from the cloud dashboard or Telegram. Get notified before your agent deletes files, sends emails, or runs dangerous commands.
Diagnostics UI
Toggle openclaw.json flags from the dashboard without editing JSON. See diagnostic events, service health, and debug tools in one place.
Rate limit monitor
Track Claude and Gemini API rate limit windows in real time. See how close you are to hitting limits and optimize request patterns.