ClawMetry is a free, open-source dashboard for monitoring AI agents. It provides real-time tracking of token costs, cron jobs, sub-agents, and memory.
ClawMetry is an open-source real-time observability dashboard purpose-built for AI agents running on the OpenClaw runtime. It is designed for developers, engineers, and teams who build and deploy AI agents and need instant visibility into what those agents are actually doing. With over 477,000 installs across 126 countries, it supports a wide range of runtimes including OpenClaw, NVIDIA NemoClaw, NanoClaw, PicoClaw, Claude Code, Codex, Cursor, Aider, Goose, and Hermes. The core value proposition is eliminating the black box: instead of hoping agents are working correctly, users gain a live, glanceable view of every sub-agent, tool call, token cost, and session log. ClawMetry runs locally, requires no configuration, and is free under the MIT license, making it accessible to individual developers and large fleets alike.
AI developers face a critical blind spot: agents spawn sub-agents, burn tokens, call tools, and make decisions behind the scenes, but there is no built-in way to see what is happening in real time. Costs can spiral unnoticed when a single agent enters a loop or calls an expensive model repeatedly. Teams often only discover runaway expenses when the monthly invoice arrives. In production, a single misbehaving agent can burn through thousands of dollars in minutes, and without observability, the cause remains invisible until it is too late. ClawMetry solves this by surfacing every detail—token counts, cost per session, sub-agent spawn trees, and tool call logs—so users can catch problems the moment they start, not after the damage is done.
The live flow visualization shows the entire decision path an agent takes, step by step, in real time. When an agent spawns a sub-agent, the spawn tree appears instantly on the dashboard, displaying each sub-agent’s status, tool calls, token usage, and progress. This feature is critical for detecting runaway loops: the dashboard flags an agent that is looping with messages like 'codex looping, 38 tool calls, no progress' and alerts the user before costs blow the budget. Users can stop an errant agent in one tap from the dashboard, their phone, or the optional desk device, giving them immediate control over their AI fleet. The desk device, a $49 hardware companion, provides a dedicated glanceable screen for the entire agent fleet, making it easy to monitor even without opening a browser.
ClawMetry provides granular token cost tracking across every dimension: per agent, per session, per model, and per tool. The dashboard displays tokens in, tokens out, cache hits, response times, and cost per call—all updated as agents run. Teams can see exactly which agent is spending the most, which model is the most expensive, and which tool calls are driving up costs. This level of detail lets users optimize their agent workflows without changing models: for example, a user might discover one agent is using three times more tokens than necessary and adjust its prompt or tool usage accordingly, cutting the AI bill significantly. The cost tracking also helps with budgeting and forecasting, as users can set alerts for when a session exceeds a predefined cost threshold.
Publisher
A
admin
Launch Date2026-02-18
Platformdesktop
Pricingfree
Domain Authority
Ahrefs DR24/100