Open on ProductHunt

Claude Usage Tracker

See Claude costs by project, across every AI tool

ProductivityOpen SourceDeveloper Tools
4votes1commentsLaunched 54d agoWebsite

By Thanh Thien

Claude-Zusammenfassung

Claude Usage Tracker aggregiert die lokalen Nutzungsdaten verschiedener KI-Entwicklungstools wie Cursor, Windsurf oder Claude Code und ordnet sie nach Projekten und Verzeichnissen. Für jemanden, der KI-Tools im Consulting- oder Coaching-Kontext einsetzt, ermöglicht das eine transparente Kostenzuordnung pro Kundenprojekt oder Workshop – ohne Schätzungen. Das Tool läuft lokal ohne Cloud-Anbindung, was für vertrauliche Beratungsmandate ein relevanter Vorteil ist. Die Kommentarlage ist dünn und sachlich-technisch, der Gründer erklärt selbst den Kontext, externe Stimmen fehlen weitgehend – kein erkennbarer Community-Buzz.

About

Claude Usage Tracker now has Project Mode: group local Claude sessions by working directory and see which projects burn your budget across Claude Code, Cursor, Windsurf, Aider, Cline, Roo Code, Continue.dev, Claude Desktop, OpenClaw, and more. Expand each project into sessions, filter by tool/model/date/cost, and compare API-equivalent spend. Local-first: no accounts, cloud, or telemetry.

Top comments1 of 1

  • [REDACTED]·54d ago0
    Hey Product Hunt - I launched Claude Usage Tracker here a couple of months ago with one simple promise: show your total Claude spend across all the tools you use. The most useful feedback from that launch was also the obvious next question: "But which project is costing me the most?" That was hard to answer because Claude Code, Cursor, Windsurf, Aider, Cline, Roo Code, Continue.dev, Claude Desktop, and other tools each keep their own local session data. A project can get touched by multiple tools, and each tool only sees its own slice. So I shipped Project Mode. What changed: - Costs are grouped by working directory, so one project gets one real total across tools - Projects are sorted by total spend - Each project expands into the underlying sessions - Existing filters still work: source, model, date range, and minimum cost - Token pricing includes input, output, cache read, and cache write - Export/import lets you bring data from another machine into the same local dashboard Everything still stays local. No account, no cloud sync, no telemetry. The source is MIT licensed and free to build yourself. There is also a $9 signed macOS build if you want the easiest install and update flow.