A project needs three things before Build Station will run any tasks for it: a working directory, at least one enabled agent, and the project-level enable toggle flipped on.
Pick the project
The main window lists every project your account has access to on this Meshly Build instance. Click one to open its detail panel.
Note: Where does this list come from?
The list is fetched from the backend on launch and refreshed on each resync. Build Station doesn't keep a local cache that can drift — projects deleted on the server disappear here, projects added on the server show up after the next sync.
Set the working directory
Build Station needs to know where the local checkout of the repo lives — the folder you'd cd into to work on the project by hand.
1. Click Set working directory
In the project detail panel. A native folder picker opens.
2. Pick the folder
Should be a real directory on disk. If the project has a git remote, the folder should be the cloned repo — Build Station does NOT clone for you.
3. Verify
Build Station checks the path exists and is a directory before accepting. If it's not, you get a red error and the field stays empty.
Warning: Build Station does not git-clone for you
If you haven't cloned the repo yet, do that first (
git clone <repo>in your terminal), then point Build Station at the clone. We don't auto-clone because we don't want to guess your authentication setup, target directory, or branch.
Enable agents
Each project has one or more agents — defined on the dashboard side. They appear in the project detail panel with a toggle each. Flip on the ones you want this Build Station to run.
See Platform agents and project agents for what the difference is and which to enable.
Flip the project on
The top-level Enabled toggle. Until this is on, Build Station ignores the project entirely — the agents won't get SSE listeners and the queue poller won't fetch tasks for them.
You can disable a project temporarily (e.g. while you're refactoring locally and don't want agents touching anything) without losing the working-dir or per-agent settings. They come right back when you re-enable.
Per-station, not per-account
The project list IS account-wide (it comes from the backend), but the per-project settings on Build Station are local to this machine. The working directory, enabled state, per-agent enables — none of that syncs to other Build Stations you might have signed in. That's intentional: different machines often have different repo layouts.
Disconnecting a project
To stop using a project on this machine entirely:
1. Disable
Flip the top-level toggle off. Build Station stops listening immediately. Any in-flight task finishes; nothing new gets picked up.
2. (optional) Clear the working directory
Click the working-directory field and remove it. Build Station forgets the path; if you re-enable later you'll be asked to pick again.
Next
-
Enabling and disabling agents: Per-agent toggles, Platform vs explicit agents, why you'd disable some.
-
Worktree mode: Let multiple agents work on the same project in parallel via isolated git worktrees.