Machine not reporting?

You enrolled a machine but it isn't appearing in the console, or it stopped. Work down this list — most cases are one of the first three.

1. Node is installed and recent

The agent needs Node 20+. Check:

node --version

If it's missing or older, install/upgrade Node and re-run the enroll one-liner.

2. The agent actually enrolled

Look for the agent's home directory — ~/.tokentracker/ (or wherever TB_HOME points). It should contain agent state and a WHAT-IS-COLLECTED.txt disclosure receipt. If it isn't there, the install didn't complete — re-run the install one-liner with your ingest key.

3. The scheduled run exists (and has run once)

Enrollment schedules a daily run — Task Scheduler (Windows), launchd (macOS) or systemd (Linux). A machine reports after its first run, so right after enrolling you may simply be waiting for it. To confirm connectivity now, re-run the enroll one-liner: enrolling round-trips to the console, so a successful enroll means the key and network are good.

4. Outbound HTTPS isn't blocked

The agent posts to the console over HTTPS. A corporate proxy or firewall can block it. The agent is offline-tolerant — it queues rollups to a local outbox and ships them when connectivity returns — so a machine behind a blocked network will appear once it can reach the internet (or once the proxy allows the console host).

5. The ingest key matches your org

The ingest key ties a machine to your tenant. If you used an old or wrong key, mint a fresh one in Admin → Keys (or Onboarding) and re-enroll. Keys are stored only as a hash, so mint a new one freely.

6. There's actually usage to report

A machine with no tracked AI-tool activity yet has nothing to ship. Use a tracked tool (Claude Code, Codex, …) on it, then wait for the next run.

Still stuck after these? Export what the agent sees and email support — include your OS, node --version, and whether ~/.tokentracker/ exists.