App Build
Build the Meadow desktop app from source and launch the built app in test mode so you can verify it starts and runs correctly.
Important: The shell working directory persists between Bash tool calls, so relative cd commands cause doubled-path errors. As the very first Bash command, resolve the repo root and store it in a variable:
REPO_ROOT=$(git rev-parse --show-toplevel)Then use $REPO_ROOT in all subsequent commands. Never use bare relative paths like cd app/hosts/desktop.
Step 1: Check for a running Meadow app
The build launches a freshly-built Meadow app in test mode. If an existing Meadow app is already running, macOS's single-instance behavior kills the newly launched instance before it finishes starting, and the verification step fails with a confusing startup timeout.
pgrep -f "Meadow.app/Contents/MacOS/Meadow" || trueIf this returns any PIDs, stop immediately and tell the user:
"app-build aborted: a Meadow app is already running. Please quit Meadow (cmd+Q) before running /app-build again."
Do not attempt to kill it yourself — the user may have unsaved state.
Step 2: Build and launch
cd "$REPO_ROOT/app/hosts/desktop" && ./build-and-test.shThis builds all components (rust binaries, backend, frontend, electron main), packages the app, mounts the result, and launches Meadow with --test-mode. The script waits for === STARTUP_COMPLETE === in the test log and then blocks so you can interact with the running app. Press Ctrl+C to stop — the script unmounts the volume and kills the launched process on exit.
Step 3: Summary
When the user stops the session, report:
- Whether backend, frontend, and main-window startup markers all appeared in the test log
- The built app location under
app/hosts/desktop/build/