Menu
Home/How it works

The operational loop

Intent in. Evidence out.

DotBridge is intent-based. An agent says what it is trying to do. DotBridge answers with the next lawful step as a ready-made, typed request—then keeps the work oriented all the way to a receipt.

Intent to receipt loopA four-stage loop shows intent moving to next step, then to a door, then to a receipt, which reorients the next intent.01 / DECLAREIntentwhat am I trying to do?02 / ORIENTNext steptyped request, ready-made03 / ROUTEDoorlocal or owner-approvedremote04 / CLOSEReceiptevidence for completion
Every loop makes the next state legibleIntent is not permission

Locality first

One bridge, two doors.

The door follows where the agent is. On-device clients use a local door. Cloud agents use an owner-approved remote door, with owner windows and Touch ID in the loop.

On-device client

Local door

The client and the work are on the same machine. The route stays local to the owner’s computer and repositories.

  • Local route
  • Owner’s computer
  • Read, search, write with custody
Cloud agent

Remote door

The route is not implied by the chat. The owner approves the remote door through defined owner windows with Touch ID.

  • Owner-approved route
  • Remote-capable chat agent
  • Custody remains visible

From change to closure

Contract. Patch. Gate. Land.

For repository work, DotBridge frames the change as a sequence that can be inspected and closed. Approved tests belong on the route; the gate is the point that decides whether the work can land.

contract

Defines the intended change in a form the next step can use.

scope
patch

Holds the proposed repository change under custody.

change
gate

Places approved tests and required checks before landing.

evidence
land

Records the completed result through a receipt.

closure