TokenLens — AI Coding Insights

Monitor your Copilot, Cursor, and AI coding tool usage. Find duplicate completions, track token spend, and prove ROI.

TokenLens passively monitors your AI coding assistant and surfaces insights you can't get anywhere else — all without leaving your editor.

Features

Completion Tracking

How many completions you accept vs dismiss, broken down by language and file type. See which languages get the best AI suggestions and where you might need better context.

Token Spend Estimation

Estimated cost of your AI coding tool usage with monthly and annual projections. Know exactly what Copilot is costing your team per developer.

Duplicate Detection

Finds boilerplate patterns generated repeatedly across your codebase — import blocks, error handlers, CRUD patterns, and more. Uses a 3-layer detection pipeline (exact hash, normalized match, template patterns).

Status Bar Widget

Live token spend and acceptance rate right in your editor. Click to open the full dashboard.

Export & Team Analysis

Export session data as JSON and upload to tokenlens.co for team-wide analysis, waste detection across all developers, and savings reports.

Why It Matters

A 50-seat Copilot Business plan costs $11,400/year. Most teams have:

• 30-40% inactive seats (no completions in weeks)
• 45%+ duplicate completions (same boilerplate regenerated)
• 22% average acceptance rate (most suggestions rejected)

TokenLens gives engineering leaders the data to justify, optimize, or cut AI coding tool spend.

Getting Started

1. Install the extension
2. Start coding — TokenLens monitors automatically in the background
3. Click the status bar widget or run TokenLens: Show Dashboard from the command palette
4. Export data to upload to tokenlens.co for team-wide analysis

Commands

Settings

Cloud Sync (Optional)

With a TokenLens Growth plan, sync anonymized usage data to the web dashboard for team-wide visibility. Only counts and metrics are synced — never your actual code or prompts.

{
  "tokenlens.syncToCloud": true,
  "tokenlens.apiKey": "your-api-key-from-tokenlens.co"
}

Privacy

• Local-first — All data stays on your machine by default
• No code sent — We never read or transmit your source code
• Opt-in sync — Cloud features are off unless you explicitly enable them
• Anonymous metrics — Sync sends only counts (completions, tokens, languages) — never text

Supported AI Coding Tools

Works with any AI coding assistant that provides inline completions in VS Code:

• GitHub Copilot
• Cursor
• Codeium / Windsurf
• Tabnine
• Amazon Q Developer
• Any other IntelliSense-based completion provider

Links

• TokenLens Web Dashboard
• Copilot ROI Calculator
• Documentation
• Report Issues

License

MIT — PrAry Soft