Customizable timeout for worktree creation

[Clindbergh]

What is the reason that the timeout for the worktree creation script exists? Would it be possible to change this? My script should install dependencies, setup a db etc. - 60 seconds might not be enough for that.

[asheshgoplani]

Shipped in v1.7.65. Thanks @Clindbergh — legit use case (dep install + DB setup in worktree hook > 60s).

New config:

[worktree]
setup_timeout_seconds = 600

Defaults to 60s for backward compat. Passes through to the exec.CommandContext timeout in the worktree-setup path.

Release: https://github.com/asheshgoplani/agent-deck/releases/tag/v1.7.65
PR: #727

[Clindbergh]

Do I see the output of that script somewhere so I can determine if it failed? Or not possible?

[asheshgoplani]

Depends on the path you used.

• CLI (agent-deck launch …, agent-deck add … --worktree, agent-deck session worktree-setup) — stdout/stderr stream to your terminal. Exit code, error output, the lot. git/setup.go:33 RunWorktreeSetupScript takes explicit stdout, stderr io.Writer parameters; the CLI wires both to os.Stdout/os.Stderr.

• TUI (creating from the New Session dialog) — same internals get called but the writers are connected to in-process buffers that arent surfaced. So if the script fails on the TUI path, no visible signal.

That second case is the gap you just filed as #768 ("Setup script needs user-visible progress + success/failure indication before claude starts"). Right home for follow-up there.

Workaround until #768 lands: when a TUI-created worktree behaves like setup didnt run, drop to a shell and re-run manually:

cd ~/.agent-deck/worktrees/<your-worktree>
bash .agent-deck/worktree-setup.sh

The exit code + output will show what failed.