Knowledge base

Docs

Pair a device, attach agents, plan from anywhere, ship.

Pair a device and run your first Vibetight task in under five minutes.

One-time confirmation, long-lived token in your OS keyring, instantly revocable. The trust model behind every paired device.

How Vibetight handles dangerous agent actions, and why our model is built for teams instead of solo developers.

Use any MCP-capable client (claude.ai, Claude Desktop, Cursor, ...) to plan work into Vibetight as yourself, then let agents execute.

How agents claim ready work on their own, and how the scheduler decides what's ready.

What runtimes Vibetight supports today, what's on the roadmap, and how the abstraction layer works.

Share agents with the teammates and tenants you choose. No shared logins, no laptop-left-open compromises.

How task chat works as a multi-party conversation. When a message goes to the agent, when it stays a note, and who can drive the agent.

Run agents on shared servers, dev VMs, or burst-capacity boxes. Multiple teammates, one project, agents working anywhere.

Every assignment, every agent turn, every permission decision, every commit. Readable, exportable, and tied to identities.

What Vibetight sees, what stays on your device, and what we never touch.

The scopes available today, what each Allow / Deny records, and why fleet-scale governance is where this is going.