Know what comes next
The developer says the feature is implemented and asks what must happen before trying it. dev-lead@factory checks the team’s recorded process, rather than relying on its memory of the wider sequence.
What happens
Locate the current step
rig workflow continue reports the instance’s current position: the build packet is owned by dev-lead@factory. Despite its name, this call inspects the workflow; it does not advance it.
Read the purpose and sequence
The agent reads the authored guidance: build, then an independent combined check, then a human tryout. It also reads the proof file, which says the interface and behavior passed separately but still need to be checked together.
Make the handoff
The lead decides the combined check is next and explicitly calls rig workflow project with the current packet, handoff exit and evidence reference. OpenRig closes that packet and creates the next task for qa-test@factory.
Give the next owner enough direction
qa-test@factory reads and claims the new task. The lead tells the developer that the independent check comes before the tryout.
What OpenRig does
OpenRig keeps the workflow’s position and authored sequence available and records the explicit transition to the next owner. The read calls and the command that advances the workflow are separate actions.
What Claude or Codex does
The orchestrating agent interprets the guidance, examines evidence and decides how to proceed. The workflow provides structure for a long process while the agent still exercises judgment.
When this helps
An orchestrator can lose track of what comes next as a project grows. A recorded workflow lets it recover its place and the purpose of the next step without asking the human to reconstruct the process.