Analysis updated 2026-05-18
Track token usage and cost across all requests your app sends to Anthropic or OpenAI
See a rebuilt view of full chat sessions from otherwise stateless API calls
Identify which specific tool calls inside an agent are consuming the most tokens
| tryrekon/rekon | 4ssh1/mine-sweeper | achawla19/intuitcode-extension | |
|---|---|---|---|
| Stars | 2 | 2 | 2 |
| Language | TypeScript | TypeScript | TypeScript |
| Last pushed | — | 2026-01-31 | — |
| Maintenance | — | Maintained | — |
| Setup difficulty | moderate | moderate | moderate |
| Complexity | 3/5 | 2/5 | 4/5 |
| Audience | developer | vibe coder | developer |
Figures from each repo's GitHub metadata at analysis time.
Requires a Cloudflare account with D1 and Workers for your own deployment, or you can use the hosted tryrekon.com instance instead.
Rekon, also called Token Profiler, is a transparent proxy that sits between your app and the Anthropic or OpenAI APIs, recording how many tokens each request uses and showing that usage in a dashboard. You point your existing AI client at Rekon instead of directly at Anthropic or OpenAI, keep coding exactly as before, and Rekon quietly logs per request token counts, rebuilds full conversation sessions, attributes usage down to individual tools, and estimates the dollar cost of each request based on published list prices. It is built on Cloudflare Workers for the proxy itself, a Cloudflare D1 database for storage, the Hono web framework, and React for the dashboard front end. A hosted version is already running at tryrekon.com, so you can use it without deploying your own copy. Because requests are streamed straight through to the real API and recording happens outside the critical path, the proxy is designed to add no noticeable delay. It also never stores your API keys: your authentication headers are simply forwarded along to the real API on each request, so usage is always billed to your own account. To use it, you create a system in the dashboard, which gives you a unique proxy URL, then set that URL as your ANTHROPIC_BASE_URL or OPENAI_BASE_URL environment variable in whatever client you already use. The system id in that URL also acts as a secret key, so only requests carrying a known id get proxied and forwarded. Running your own deployment involves cloning the repository, installing dependencies, setting up local secrets including an OAuth provider like Google or GitHub, running database migrations, and starting a local development server, with a similar but separate process documented for a full Cloudflare deployment. This tool fits developers who already build with the Anthropic or OpenAI APIs and want visibility into exactly how many tokens their app, or the different tools inside it, are consuming and what that costs, without changing how their client talks to the API. The project is released under the Apache 2.0 license and describes itself as fully open source, with a hosted version as its commercial offering rather than a paid version of the code.
A transparent proxy for the Anthropic and OpenAI APIs that logs token usage, rebuilds conversations, and estimates cost per request.
Mainly TypeScript. The stack also includes TypeScript, Cloudflare Workers, D1.
Use freely for any purpose, including commercial use, as long as you keep the copyright notice and state changes you made.
Setup difficulty is rated moderate, with roughly 30min to a first successful run.
Mainly developer.
This repo across BitVibe Labs
Verify against the repo before relying on details.