50-seat Copilot Business costs $11,400/year. GitHub shows you seats used — TokenLens shows you whether those seats are generating real value or burning budget.
Also works with Cursor · Codeium · Windsurf · Any IDE telemetry
Engineering leaders spend thousands on AI coding tools with zero visibility into actual productivity impact.
Developers who activated Copilot but haven't generated a completion in weeks. You're paying $19–39/month for nothing.
of seats are inactive at most orgs
The same boilerplate — imports, error handlers, CRUD patterns — generated thousands of times across repos, burning through token limits.
of completions are near-duplicates
Developers reject most suggestions but still consume tokens. Without metrics, you can't tell who needs better prompting training.
average acceptance rate (industry)
Start with a simple CSV export. Go deeper with our IDE extension.
Go to your GitHub org settings → Copilot → Usage → Export CSV. Drop the file into TokenLens. Takes 30 seconds.
Organization → Settings → Copilot → Export
✓ Available now · Free tier
If you route Copilot/Cursor through a proxy (LiteLLM, Helicone, custom gateway), export the request logs as JSON or JSONL.
litellm, helicone, custom proxy
✓ Available now · Starter tier
Install our lightweight extension. It passively monitors completions, tokens, and acceptance rates — then surfaces insights in a local dashboard.
ext install tokenlens.copilot-insights
✓ Available now · Growth tier
Every metric engineering leadership needs to justify — or cut — AI coding tool spend.
Which developers are active, inactive, or low-usage. Projected savings from reallocating idle seats. Drill down by team, editor, and language.
Find boilerplate patterns generated repeatedly across your org. Candidates for shared snippets, templates, or a custom context library.
How much each completion costs by language, editor, and developer. Identify which teams get the best value from their AI coding tools.
Which developers accept suggestions vs dismiss them. Low acceptance rates signal a training opportunity — not necessarily a tool problem.
Is adoption growing or plateauing? Track usage trends over time to catch engagement drops before renewals.
Based on your actual usage data: what your Copilot spend delivers per developer per month, and where optimization would have the highest impact.
The TokenLens VS Code extension passively monitors Copilot, Cursor, and other AI coding tools. No proxies needed. Everything stays local.
Every tier includes Copilot analysis. Upgrade for deeper insights and team features.
Upload your GitHub Copilot usage export. See exactly which seats are delivering value and which are burning money.
Free forever. No credit card. 30-second setup.