See your OpenClaw agents' costs, activity & memory live
ClawMetry is a free, open-source observability dashboard for OpenClaw AI agents. Think Grafana, but purpose-built for AI. One command install (pip install clawmetry), zero config. Monitor token costs, sub-agent activity, cron jobs, memory changes, and session history. All in real-time with a beautiful live flow visualization. Works on macOS, Linux, Windows, even Raspberry Pi
ClawMetry provides real-time token tracking, per-session cost attribution, and budget alerts. Users can see token usage per model, per tool, and per session, and set budget limits with alerts to catch runaway agents before they incur expensive mistakes. Pro plan adds per-run waste flags to identify inefficient token consumption.
ClawMetry provides session history with full transcripts, timeline views, tool call logs, and error triage. Users can replay what an agent did, identify where it got stuck, and see cost breakdowns per run. Pro plan adds error mute/resolve workflows and anomaly detection for spend and error spikes.
ClawMetry's live flow visualization shows animated, color-coded agent activity as it happens—message channels, LLM requests, tool calls, and responses. Users can see sub-agent spawning trees, tool execution sequences, and error states instantly, eliminating the 'black box' problem of agent execution.
Pro and Enterprise plans include approval queues to block tool calls by policy, tool catalog policies with pre-execution gates, and custom webhooks for PagerDuty/OpsGenie integration. Users can define which tools agents can call and require approval before execution, with audit logging of all decisions.
Starter and Pro plans support unlimited nodes and multi-runtime monitoring (OpenClaw, Claude Code, Codex, Cursor, Aider, Goose, opencode, Qwen, Hermes). Cloud sync with E2E encryption enables fleet-wide visibility. Users can view all agents in one dashboard with per-runtime health timelines and cost attribution.