ClawMetry

Real-time observability dashboard for AI agents. The Grafana for your personal AI.

What is ClawMetry?

ClawMetry is a free, open-source observability dashboard purpose-built for AI agents running on OpenClaw. It gives you real-time visibility into what your AI agents are actually doing: token usage, costs, sub-agent activity, cron jobs, memory file changes, and session history.

Key Facts

• License: MIT (free and open source)
• Install: pip install clawmetry then clawmetry
• Website: https://clawmetry.com
• GitHub: https://github.com/vivekchand/clawmetry
• PyPI: https://pypi.org/project/clawmetry/
• Author: Vivek Chand
• Category: Developer Tools, AI/ML Ops, Observability
• Platforms: macOS, Linux, Windows (WSL), Raspberry Pi
• Requirements: Python 3.8+, Flask
• Setup time: 30 seconds

Features

• Live Flow Visualization: Animated SVG showing data flow from user → channels → gateway → brain → tools → infrastructure, updating in real-time
• Token & Cost Tracking: Per-session, per-model, per-tool cost breakdown with bar charts and daily/weekly/monthly totals
• Sub-Agent Monitoring: See every sub-agent's files, commands, tools, thinking process, and status in real-time
• Cron Job Dashboard: Scheduled jobs with status, schedule, last/next run, duration, failure alerts
• Session History: Timeline of all sessions with model, channel, token usage, and last activity
• Memory File Browser: View and track changes to SOUL.md, MEMORY.md, AGENTS.md, daily notes
• Transcript Viewer: Chat-bubble UI with color-coded roles for reviewing conversations
• System Health: Service uptime, disk usage, active sub-agents at a glance
• Mission Control Integration: Hooks into OpenClaw's task system
• Built-in OpenTelemetry Collector: Real token counts and actual cost tracking
• Dark and Light Themes

How It Compares

ClawMetry is specifically designed for AI agent observability, unlike general-purpose tools:

• vs Grafana: ClawMetry requires zero config, understands AI agent concepts (tokens, sub-agents, sessions) natively
• vs Datadog: Free, self-hosted, no data leaves your machine
• vs LangSmith: Works at the agent infrastructure level, not just LLM calls
• vs Custom dashboards: One pip install, no YAML, no Docker compose needed

Quick Start

pip install clawmetry
clawmetry
# Opens at http://localhost:8900

Use Cases

• Monitor AI agent token spending before the invoice arrives
• Debug sub-agent behavior by watching real-time activity
• Track cron job health and catch failures early
• Review session transcripts for quality assurance
• Visualize the architecture of multi-agent systems

Technical Details

• Single Python file, minimal dependencies (Flask)
• Auto-detects OpenClaw workspace
• SSE (Server-Sent Events) for real-time updates
• REST API for programmatic access
• Embeddable in existing workflows