OpenRig
← All examples

Ask the specialist

The developer wants to connect a new app feature to the backend. dev-ui@factory handles the interface; dev-app@factory maintains the backend and knows why the team chose its current connection method.

What happens

  1. Recognize whose knowledge matters

    dev-ui@factory could inspect the code, but that alone might miss a recent decision. It asks the teammate responsible for the backend.

  2. Send the question to that terminal

    The command rig send dev-app@factory delivers the question into the other agent’s prompt. The incoming message identifies who sent it. dev-app@factory can respond using its own project context and tools.

  3. Check the source and reply

    dev-app@factory reads notes/integration.md with an ordinary shell command. The note explains that the shared client was updated last month to handle reconnecting. It sends that answer back to dev-ui@factory.

  4. Use the answer

    dev-ui@factory chooses the shared client, preserving work the team has already done. The exchange answers a question; it does not create a new implementation assignment.

What OpenRig does

OpenRig resolves the named seat to its terminal and delivers the message. The reply uses the same mechanism in the other direction. A rig send message does not by itself create a tracked queue task.

What Claude or Codex does

The agents decide what to ask, which project file to inspect and how to apply the answer. The cat command is ordinary file reading by the coding harness; it is not an OpenRig command.

When this helps

Use this when another specialist has the context you need and the immediate outcome is an answer. Reserve a tracked task for work that needs an owner and later follow-up.