How do I monitor PicoClaw agents?

Install ClawMetry (pip install clawmetry). The daemon auto-detects ~/.picoclaw/workspace and renders every session, tool call, and provider-attributed cost in one dashboard at http://localhost:8900.

Why do some PicoClaw sessions show $0 without ClawMetry?

PicoClaw is token-blind on providers that do not self-report usage (the message store carries content but not tokens). ClawMetry's hosted-VM log ingestion (CLAWMETRY_VM_USAGE_LOG, gated behind CLAWMETRY_VM_USAGE_RUNTIME=picoclaw so it never double-counts self-reporting runtimes) attributes those calls at ingest time, so the Cost tab shows a real number instead of $0.

Does ClawMetry work with PicoClaw's cron scheduler?

Yes. Scheduled PicoClaw jobs are surfaced on the Cron tab with last-run, next-run, duration, and failure alerts.

Can I catch a PicoClaw runaway loop?

Yes. Per-session token growth and message-repeat signals trigger the approval queue; the kill switch (Pro) stops the session before the next call.

How much does ClawMetry cost for PicoClaw?

Free forever on your own machine. Cloud starts at $9/node/mo (Starter) or $19/node/mo (Pro).

Observability and cost tracking for PicoClaw agents

See every PicoClaw session, tool call, and per-provider token cost as it happens. ClawMetry reads PicoClaw's workspace under ~/.picoclaw/workspace and its hosted-VM per-call LLM log, so sessions on providers PicoClaw itself does not report tokens for still get cost attribution.

Supported runtime Reads flat providers.Message JSONL per session Read-only Free trial

What ClawMetry sees on PicoClaw

ClawMetry's PicoClaw adapter opens the store PicoClaw already writes (~/.picoclaw/workspace (plus the hosted-VM CLAWMETRY_VM_USAGE_LOG when PicoClaw runs in a managed VM)) in read-only mode and surfaces:

Sessions: started at, last activity, model, message count, per-session token totals.
Live activity: every tool call as it lands, with args and result. Turn-level tokens split.
Per-turn tokens: Per-provider usage extracted from PicoClaw's provider messages plus a loopback LLM shim log for hosted VMs, so managed-VM sessions with no self-reported tokens still get cost attribution.
Cost per session, per model, per hour: cache-aware where the model supports caching.
Per-provider cost attribution (PicoClaw itself is token-blind on many providers): the PicoClaw-specific signal ClawMetry surfaces that generic dashboards do not.
Approval and kill switch: watch for token-maxing signals (message repeat, no-progress heuristic, cost-per-hour threshold), pause or kill before the next call.

Source: the adapter is at clawmetry_pro/adapters/picoclaw.py. Every parse rule and every SQL SELECT is in that file.

Enable it in about two minutes

1. Install ClawMetry

Pick the line that matches your OS. Windows users: paste the install.cmd line, not the shell one.

# macOS or Linux
curl -fsSL https://clawmetry.com/install.sh | bash

# Windows (cmd.exe or PowerShell)
curl -fsSL https://clawmetry.com/install.cmd -o install.cmd && install.cmd

2. Run the daemon

clawmetry

The daemon auto-detects PicoClaw's local store and starts reading immediately.

3. Open the dashboard

Open http://localhost:8900. Switch the runtime chip to PicoClaw.

Start the free trial →

The one thing this catches that PicoClaw alone does not

Left alone, PicoClaw's failure mode is a session that reads $0 on the Cost tab because the underlying provider does not report tokens. ClawMetry watches for it: cost per hour on the PicoClaw runtime, message-repeat detection, and per-session token-growth curves. Set a threshold, and the approval queue pauses before the next call. Kill switch (Pro) stops it.

All 19 runtimes ClawMetry observes

PicoClaw is one of 19. Free on every plan: OpenClaw, NVIDIA NemoClaw. Paid on Cloud Pro or Self-Hosted Pro: the rest.

OpenClaw NVIDIA NemoClaw NanoClaw PicoClaw (this page) Claude Code Codex Cursor Aider Goose opencode Qwen Code Hermes Pi Deep Agents n8n Antigravity GitHub Copilot Grok QM