Run the server
Reach, back up, upgrade and run the Kraft server.
Added in 2.1
Each guide takes one task from start to finish.
In this section
- Remote access: reach the board from a phone or another machine, over Tailscale, your local network or a tunnel, with a password.
- Put a proxy in front of Kraft: let a tunnel or reverse proxy on this machine pass webhooks and scripts to a loopback-bound Kraft.
- Restart Kraft without ending running agents: pause the items that have an agent running, restart the server, and resume them.
- Back up and restore the database: copy the database while the server runs, restore it, and use the copy Kraft makes before an upgrade.
- Logs and disk space: find the logs, and reclaim the space worktrees use.
- Run Kraft as a service: start Kraft at login, and run a second instance.
- Upgrade Kraft: back up, upgrade and check an installed Kraft, pin a version, and roll back to an earlier 2.x release.
- Upgrade your configuration: take new shipped chains, library tasks and policy after an upgrade without losing your edits.
- Upgrade from 1.4: what 2.0 changes from 1.4, and the steps to update from it.
- Upgrade from 0.x or a release candidate: move from a 1.5.0 release candidate, an install from before the
kraft-sdlcrename, or a 0.x home with the old template configuration. - Roll back to 1.4: what a rollback does to the database and config, and the steps from 2.0 back to 1.4.
- Run items in your own checkout: let a repo's work items run in its own checkout instead of a worktree each, and clear a
checkoutstop.
Give a worker its login
Give the agent CLI a worker runs on, whether Claude Code, Codex, Cursor, OpenCode, Amp, Antigravity or Gemini CLI, a login it can use headless, and check that the worker reaches it.
Remote access
Approve or reject a gate from a phone, over Tailscale, a tunnel or a reverse proxy you run, using the same password auth as the local board, get back if Kraft does not restart, and rotate the token.