Skip to main content
A run script is a shell command saved on a repo. When you press Play in the run sidebar, p0 spawns the script in a PTY rooted at the repo’s checkout.

​
Configuring a script

Each repo row in the repo manager has a run-script editor (it’s also reachable from the run sidebar’s “No run script configured” empty state). The dialog is titled Run Script and contains:
  • A textarea for the command. Placeholder is e.g. bun run dev. Any shell command goes here — single line or multi-line.
  • A button labeled Add path-specific run command for monorepos. Each entry is a sub-path inside the repo (e.g. apps/api) with its own script. The path input autocompletes from the repo’s directory listing.
Saving the dialog with an empty textarea clears the root script. Toast: Run script cleared for .

​
How p0 invokes it

When you press Play, p0:
  1. Looks up the saved script for that repo (or sub-path).
  2. Writes it to a temp file — .sh with a #!/bin/bash shebang on macOS/Linux, .cmd on Windows.
  3. Spawns a PTY with the working directory set to {worktree}/repos/{repo-name} (plus the sub-path, if any) and runs the temp file (exec bash <path> on Unix, call "<path>" & exit /B on Windows).
  4. Streams stdout/stderr into the xterm.js pane in the run sidebar.
The PTY accepts stdin, so prompts like “Ok to proceed? (y)” work — type into the pane and the keystrokes go to the process.

​
Multiple scripts per repo

There’s one root script per repo. On top of that, you can add any number of path-specific scripts for sub-directories — useful for monorepos where apps/web and apps/api each need their own dev command. Every sub-path with a script becomes its own tab in the run sidebar, labelled {repo}/{sub-path}.

​
Lifecycle: Start, Stop, restart

The controls bar on each run-sidebar tab has a single Play/Stop button plus a status dot:
  • Gray dot — idle, not started this session.
  • Green dot — running.
  • Red dot — exited. The exit code shows on the right when it’s non-zero.
Start is idempotent — pressing Play while the process is already running is a no-op and returns the same pane. Stop is lenient — it returns success even if the process has already exited. Pressing Play after an exit resets the terminal and spawns a fresh process.