ClawMetry is a free, open-source real-time observability dashboard for OpenClaw AI agents. It shows what your agents are doing by monitoring token costs, sub-agent activity, cron jobs, memory changes, and session history.

ClawMetry is a free, open-source real-time observability dashboard designed specifically for developers and teams building with OpenClaw AI agents and other supported runtimes. Its primary purpose is to provide immediate, comprehensive visibility into the complex operations of AI agents, which often work autonomously by spawning sub-agents, calling tools, and executing tasks. The dashboard is built for engineers, founders, and operations teams who need to monitor, debug, and control their AI agent fleets to prevent budget overruns and ensure reliable performance. It serves as a mission control center, transforming the opaque inner workings of AI agents into clear, actionable data streams that can be monitored in real time from a single interface, whether on a laptop, server, or dedicated desk device.

Before tools like ClawMetry, developers working with AI agents faced significant challenges with zero visibility into their operations. Agents would spawn sub-agents, burn through tokens, and call various tools without any clear indication of what they were doing or how much they were costing. This lack of transparency meant that runaway loops could spiral costs overnight, and teams would only discover issues when invoices arrived or customers reported problems. Developers were left hoping their agents were working correctly, unable to see if an agent was stuck, which websites it was searching, or where tokens were being consumed. This problem was particularly acute with sub-agents spawning other sub-agents, leading to fast cost escalation and debugging nightmares, as there was no way to tie every token back to the specific task that spent it.

One of the first major feature groups is comprehensive cost tracking and budget protection. ClawMetry meticulously monitors token usage and costs per session, per model, and per tool, providing clear breakdowns before invoices arrive. It flags runaway loops the moment they start, such as detecting 'codex looping, 38 tool calls, no progress,' and can alert users when an agent crosses a predefined budget threshold. This allows users to stop problematic agents with a single tap from their phone, dashboard, or desk device, preventing budget overruns. The system shows the live spawn tree of sub-agents, making it easy to identify which agent or task is responsible for excessive spending, thereby giving teams the control to manage their AI bills effectively without needing to change their underlying models.

The second major feature group centers on real-time operational visibility and debugging. ClawMetry provides a live flow visualization that shows the actual decision path an agent took, step by step. It monitors channels, gateways, models, tools, and nodes, updating this information in real time as events happen. This allows users to diagnose a stuck agent in seconds, seeing exactly where it is in its workflow, what tools it is calling, and whether it is making progress. The dashboard displays cron job status, service uptime, disk usage, and active sub-agents at a glance, ensuring operational health is always visible. Session transcripts and tool call traces are logged, creating a complete audit trail for post-mortem analysis or compliance reviews.

ClawMetry

Key Features

Use Cases

Who is this for?

Comments