Skip to main content

Track Usage for Coding Tools

Track usage and costs for AI-powered coding tools like Claude Code, Roo Code, Gemini CLI, and OpenAI Codex through LiteLLM.

Monitor requests, costs, and user engagement metrics for each coding tool using User-Agent headers.

Who This Is Forโ€‹

Central AI Platform teams providing developers access to coding tools through LiteLLM. Monitor tool engagement and track individual user usage patterns.

What You Can Trackโ€‹

Summary Metricsโ€‹

  • Cost per coding tool
  • Successful requests and token usage per tool

User Engagement Metricsโ€‹

  • Daily, weekly, and monthly active users for each User-Agent

Quick Startโ€‹

1. Connect Your Coding Tool to LiteLLMโ€‹

Configure your coding tool to send requests through the LiteLLM proxy with appropriate User-Agent headers.

Setup guides:

2. Send Requests with User-Agent Headersโ€‹

Ensure your coding tool includes identifying User-Agent headers in API requests.

3. Verify Tracking in LiteLLM Logsโ€‹

Confirm LiteLLM is properly tracking requests by checking logs for the expected User-Agent values.

4. View Usage Dashboardโ€‹

Access the LiteLLM dashboard to view aggregated usage metrics and user engagement data.

Summary Metricsโ€‹

View total cost and successful requests for each coding tool.

Daily, Weekly, and Monthly Active Usersโ€‹

View active user metrics for each coding tool.

How LiteLLM Identifies Coding Toolsโ€‹

LiteLLM tracks coding tools by monitoring the User-Agent header in incoming API requests (/chat/completions, /responses, etc.). Each unique User-Agent is tracked separately for usage analytics.

Example Requestโ€‹

Example using claude-cli as the User-Agent:

curl -X POST \
-H "Content-Type: application/json" \
-H "Authorization: Bearer sk-1234" \
-H "User-Agent: claude-cli/1.0" \
-d '{"model": "claude-3-5-sonnet-latest", "messages": [{"role": "user", "content": "Hello, how are you?"}]}' \
http://localhost:4000/chat/completions