Known limitations
Rough edges are cheaper when they are visible.
MootDeck is in beta. Some parts are young, some constraints are deliberate, and some things need more field time. Hiding that would be worse than the limitations themselves.
01Beta
This is young software.
MootDeck is usable, but it is still a beta. Some flows are rough. Some terminal behaviours depend on the tool you run. Some features exist because they solved a real local problem first, not because they have been sanded into a perfect onboarding funnel.
That is not an excuse. It is the state of the product.
02Platforms
Cross-platform does not mean frictionless everywhere.
| Area | Current limitation |
|---|---|
| Windows | SmartScreen can warn on first launch until the code-signing path is fully settled. |
| Linux | AppImage support depends on the usual desktop-library reality. Glamorous, in the way plumbing is glamorous. |
| Terminal TUIs | Full-screen tools should render, but terminal programs are wonderfully inventive at finding edge cases. |
| Large graphs | One-hop ropes are deliberate. They may feel strict when a canvas grows to many agents. |
03Boundaries
The topology model is specific, not magical.
Ropes gate MootDeck's own coordination commands. They do not turn arbitrary CLI programs into sandboxes. If you start a tool that can read a repository, call a model provider, or run a script, MootDeck does not pretend that drawing a rope rewrites those permissions.
This distinction will stay visible in the docs because hiding it would be dumb. And fragile. Mostly dumb.
04Trust
Closed source for now. Free during beta.
MootDeck is closed source for now. That matters for a local developer tool. The website should say it plainly so people can make their own call.
The product is free during beta. Paid plans are not finalized. The direction is predictable pricing for teams that already coordinate multiple CLI agents, not token markup or surprise usage billing.
05Feedback
The useful questions are concrete.
- Where does coordination break down: context, handoffs, review, test status, or window management?
- Is one-hop visibility too strict once a canvas has ten or more agents?
- Which CLI agents behave badly inside a real PTY node?
- What would make the security model clearer before you trust it?
- What should be local by default that is not local enough yet?
NEXTKeep reading