Hey Product Hunt 👋 I built RepoSpend because I use AI coding tools every day, but I wanted a simple way to understand where the token usage was actually going. I care a lot about performance and cost optimisation. Whether it is latency, infra cost, or AI usage, the first step is always measurement. RepoSpend helps answer questions like: Which repo is consuming the most tokens? Which sessions are unusually expensive? Which model or provider is driving most of the usage? How much API equivalent cost would this usage represent? Where can I improve my AI assisted workflow? It is free, open source, and local first. It runs on your machine, reads local usage data, and does not upload your code or prompts. Right now it supports Codex, Claude Code, GitHub Copilot, and experimental Cursor imports. I would love feedback from developers using AI coding tools heavily, especially around what usage insights would be most useful next.

RepoSpend

Track AI coding token usage by repo, model, and tool

Track AI coding token usage by repo, model, and tool