Lean, self-hosted task planning for high-performing software teams.
A Jira alternative that helps the people doing the work — not the people watching it.
Fast, opinionated, open source.
just dev-deps # start postgres
just dev # start backend + frontendBackend runs at http://localhost:8080, frontend at http://localhost:5173.
You'll need an OIDC provider for auth. Keyline is the recommended companion — see docs/architecture.md for setup.
hivetrack/ Go backend (HTTP API)
hivetrack-ui/ Vue 3 frontend
docs/ Architecture and design documentation
justfile Task runner — run `just` to see all recipes
just dev # start everything locally
just test # run all tests (no DB needed)
just check # lint + tests — run before pushing
just install-hooks # install git pre-push hook| Doc | Contents |
|---|---|
| docs/architecture.md | System architecture, auth, deployment |
| docs/domain-model.md | Full entity definitions and relationships |
| docs/engineering-principles.md | TDD, patterns, coding conventions |
| docs/api-and-ai.md | API design, webhooks, AI integration |
Hivetrack ships an MCP server so AI assistants (Claude, Cursor, etc.) can read and write issues directly. Auth is handled automatically on first use.
Claude Code — one command installs and registers the server:
claude mcp add hivetrack -e HIVETRACK_URL=https://your-hivetrack -- npx hivetrack-mcpIf you already have the binary installed via Homebrew or Go:
claude mcp add hivetrack -e HIVETRACK_URL=https://your-hivetrack -- hivetrack-mcpOther clients — install the binary first:
npx hivetrack-mcp # npm
brew install The127/tap/hivetrack-mcp # Homebrew
go install github.com/the127/hivetrack/mcp/cmd/hivetrack-mcp@latest # GoThen add it to your client's MCP config with HIVETRACK_URL pointing at your instance. Refer to your client's documentation for the exact config format.
Backend: Go 1.24 · PostgreSQL · CQRS + mediator · IoC DI Frontend: Vue 3 · Vite · TanStack Query · Tailwind CSS v4 Auth: OIDC — Keyline, Keycloak, Authentik, or any compliant provider