GitHours Is Not git-hours
GitHours shares its name with git-hours, an open-source family of at least six independent implementations that estimate development time from git commit history — a separate lineage of tools built around a similar idea. The original, kimmobrunfeldt/git-hours, inspired forks and ports in Python and Go and a GitHub Marketplace Action. None of these projects are affiliated with GitHours, and none of them is being scored or ranked here — the point of this page is to make the distinction unambiguous before addressing the one objection every one of them shares.
“Please note that the information might not be accurate enough to be used in billing.”
“{"total": {"hours": 9959, "commits": 11470}} — From a person working 8 hours per day, it would take more than 3 years to build Bootstrap.”
GitHours works differently in ways anyone can check. Every analysis reconstructs work sessions using a configurable gap threshold — 2 hours by default — so a burst of commits separated by a short pause counts as one continuous session instead of resetting to zero. Each session opens with an explicit, disclosed 30-minute credit for the work that happens before the first commit lands — a documented assumption, not a hidden one. That methodology is published here and in GitHours' own documentation, not hidden inside a black box, and every analysis can be shared as a client-facing link so the person paying the invoice can check the same figures GitHours calculated. kimmobrunfeldt/git-hours is an earlier, different project — its own disclaimer is taken at face value here, not argued with.
What about the local tools?
Git Timetrack is a Claude Code plugin, not a hosted product or a website — it is installed with Claude Code slash commands and every report is generated through those same commands. Its own copy says it is “100% local — All data stays on your machine in plain JSON Lines. Nothing is sent anywhere.” and “Local-first by design. No accounts, no cloud, no telemetry.” Its estimation method is diff-size-aware: “Estimates work hours from commit patterns and diff sizes. Commits close together count as continuous work. Isolated commits are estimated by scope.” It is MIT-licensed and free. It is a different category of tool from GitHours — a terminal workflow, not a hosted product that produces client-facing reports — not a competitor missing features it never claimed.
gtt lives at github.com/luizbranco/gtt. Its own description reads “git time tracker is a simple program that allows you to track how much time you spend coding per day and, at the end of a period, generate an invoice.” It is a Go CLI, MIT-licensed, distributed as release binaries, generates HTML invoices, and was archived on GitHub on 2023-12-26.
How accurate is git-based time calculation?
GitHours calculates time by analyzing intervals between consecutive commits. It uses configurable gap thresholds (default: 2 hours) to handle breaks. While not perfect for every work style, it provides objective data that's far more accurate than memory-based estimates.