Hi Product Hunt 👋 - I'm Srinivas. I built Tokmeter because I was flying blind on what my AI agents actually cost. I run a lot of them - Claude Code, Codex, Cursor, Gemini, and a few more. Each one bills me in its own silo and shows me its own slice. Claude Code knows what Claude Code cost. Codex knows Codex. Nothing knew what my day cost across all of them, or which project was quietly burning the most, or whether the model I'd just switched to was actually cheaper for the work I do. So I built Tokmeter. It reads the session files these agents already write to disk — no API keys, no account, no telemetry - and turns them into one honest ledger: cost and tokens by project, model, provider, and day. It prices against a live model catalog and keeps an append-only daily history, so the numbers survive even after the agents rotate their own logs away. Why it matters now: we're all running more agents than we were six months ago, and the spend fragments faster than any single tool can show you. The bottleneck isn't capability anymore - it's seeing what it costs and where. You can't optimize what you can't see. One rule I held the whole way: it labels estimates instead of faking precision, and nothing leaves your machine - your usage is yours. It's a CLI, a macOS menu bar app, a TUI, a web dashboard, and an MCP server so an agent can check its own spend. 20+ agents, one picture. Open source (AGPL). Curious how everyone else does this - do you track what your agents cost today, or just find out at month-end? And which agent should I parse next?

Tokmeter

Token and cost totals for every AI coding agent

Token and cost totals for every AI coding agent