Use cases

A roadmap is usually a ranked list, but features don't depend on one thing each. A performance fix can serve onboarding, retention and an enterprise deal at the same time. In GoalT each of those is a parent of the fix, and the fix collects value from all of them, so the work that unblocks the most rises to the top without anyone re-scoring it by hand.

When priorities shift, you restructure the goals near the top and every value below is recomputed from the new structure, instead of someone re-ranking a list by hand.

Maintainers carry the reason for most of the codebase in their heads. A goal tree writes it down: every goal has a description and links to the files that implement it, saved as plain JSON in .goalt/tree.json inside the repository. Commit it, and a new contributor can see why a module exists and which goal it serves before touching it.

GoalT itself is Apache 2.0, needs no account and has no server, so it fits projects that can't depend on a hosted tool.

Run /goalt:startin Claude Code and Claude reads the README, package manifest, folder structure and database migrations, then builds a tree of the project's real functional areas with the files and backend artifacts behind each one. It is a fast way to get a map of a codebase you've just inherited, and to check that map against the code.

When Claude Code works through a long task, the local dashboard shows which goal it is on: every edit, read and shell command pulses the header, edited files light up the goals they belong to, and uncommitted work stays marked until it is committed. You can follow the work against the plan instead of reading a scrolling log.

• Not an issue tracker. It doesn't replace tickets, assignees or sprints; it sits above them and explains why they exist.
• Not a root-cause tool. It says why something exists, not why it broke.
• Not a hosted service. Everything runs on your machine.

Try it: install GoalT, read how the value math works, or read the code.