Spend Attribution

How much are we spending on Cursor, Copilot and Claude Code per developer?

AI Coding Tools brings Cursor, GitHub Copilot and Claude Code spend into one view per developer, next to the rest of your AI bill. Growth plan.

In the app: AI Coding Tools · Growth plan and aboveOpen in app →No account? Start free

Coding agents bill per token, and spend per developer varies by an order of magnitude. AI Coding Tools shows it per developer and per tool, for the last 7, 30 or 90 days.

Sources

ToolHow it's readPer developer?Credential
Claude Code (via proxy)Live, from calls through Cognocient AIYes, by X-Cost-User or the key's namenone (it's your proxy traffic)
Claude Code (Console)Anthropic Claude Code Analytics API, dailyYes, by email or API key name, with lines of code and sessionsAnthropic Admin API key
CursorCursor Admin API usage events, dailyYes, by emailCursor team Admin API key (Enterprise teams)
GitHub CopilotGitHub billing usage API, dailyNo: organization levelToken of an org owner or billing manager, plus the organization name

Credentials are tested by reading yesterday's usage before they're saved, then encrypted and never shown again. On connect, the last 30 days are read; after that, the last three days are re-read every day at about 04:00 UTC because vendors finalise usage late. Sync now runs it on demand.

What the numbers mean

  • Cursor cost is chargedCents per event: what Cursor charged, including its token fee. Seat fees are not in this feed.
  • Claude Code (Console) cost is Anthropic's estimated cost per model. It covers Claude Code used with Claude Console (API) accounts, not claude.ai Pro/Max/Team/Enterprise subscriptions.
  • Copilot cost is the net amount GitHub reports for Copilot products for the organization, per day. It needs GitHub's enhanced billing platform.
  • Average per developer only counts developers with per-user data.

Cap it

Usage read from vendor APIs is after the fact. To stop runaway spend, route Claude Code through the proxy and set a Key, spend group or model-group budget.

The Cursor, GitHub and Anthropic connections follow each vendor's documented API and have not yet been run against a live customer account.

On this page