Claude usage jumped? How I added project, session and subagent attribution to my usage tracker

Description and use

Claude Usage Companion now offers detailed attribution of usage—projects, models, sessions, and subagents—enabling users to quickly identify causes of token spikes.

Problem

Users could see a weekly usage bar jump but had no insight into why the spike occurred.

Outcome

The enhanced tracker gives users actionable insights into usage spikes by attributing them to specific projects, models, sessions, and subagents, enabling proactive management of Claude usage limits.

Workflow

• 1. Install the Chrome/Edge extension from the Web Store or build Firefox from source.
• 2. Open Settings → Usage on claude.ai to link the extension to your account.
• 3. (Optional) Install the Node.js daemon via npm to access local Claude Code session data.
• 4. View the popup for session and weekly bars, including forecast and anomaly alerts.
• 5. Open the full dashboard to see Overview, Activity, and Sessions tabs.
• 6. Use the Usage Inspector to identify the largest local contributor during a spike.
• 7. Review Activity for project, model, and session breakdowns and compare with previous periods.
• 8. Manage sessions via search, resume commands, export, and cleanup warnings.
• 9. Export or restore usage history and settings from the Diagnostics panel.
• 10. Provide feedback or bug reports on GitHub.

Evidence

The popup stays small on purpose. It shows the Session (5h) and Weekly bars, and says something only when something needs attention: a limit you've already hit, or a forecast that runs out early, like

Source

The full dashboard has three tabs: Overview, Activity, Sessions

Source

When an anomaly fires on Overview, there's an inspector under it. It looks at the local Claude Code sessions active during that window and shows the largest local project with its share, the model mix, and the biggest sessions

Source