Campaign Runner
A disciplined harness for bounded, multi-step AI-assisted work with explicit validation receipts and human approval at boundaries.
Prototyped
Campaign Runner turns bounded tasks, plans, execution constraints, and validation receipts into a repeatable workflow while preserving human approval at every meaningful boundary. It is evidence-bound orchestration for complex engineering campaigns.
An orchestration posture for Nodes that need to coordinate multi-step agent-assisted work with scope, proof, and clear handoffs.
Builders and teams coordinating agent-assisted work that needs scope, proof, and clear handoffs.
Campaign Runner is the public identity for work currently published as Codex Runner and codexrun. The underlying capability and compatibility surfaces remain under those names until they are renamed.
Evidence
Status reflects evidence in the corresponding source repositories at time of writing. See the research page for the broader picture.