Ask one agent where things stand
Several agents have been working on an app. The developer asks product-lead@factory a broad question: “What can I actually try today?” They want a useful answer, not a pile of task records.
What happens
Check the recorded work
The lead reads two queue tasks. feature-001 is done; setup-001, the main-screen improvement, is still in progress. The records identify the work and its owners.
Resolve what the records do not settle
The lead asks qa-test@factory whether the feature is ready to try and where the screen change stands. The specialist replies that the feature passed its checks, while the screen change is still being checked.
Bring back one answer
The lead tells the developer that the feature is ready to try today and the main-screen change is not ready yet. It combines the task records with the specialist’s answer instead of treating a task status as the whole story.
What OpenRig does
OpenRig supplies the tracked tasks and direct message route. The lead uses rig queue for recorded status and rig send for the remaining question.
What Claude or Codex does
The agent decides what to inspect, recognizes the gap in the records and turns several pieces of information into a useful reply. Claude Code and Codex can participate in the same exchange through their separate named seats.
When this helps
The developer avoids checking tickets, commits and multiple agent terminals by hand. One conversation can reach the team’s records and expertise, then return an answer framed around what the developer wants to do.