Deploy what you specified.
Not a second guess.
Field takes the rules for going live straight from the spec. The way the work ships is read from what was agreed, not written again by hand and hoped to match.
Request access →The deploy config is a second source of truth.
The spec says one thing. The deployment is set up by hand, weeks later, by someone reading the spec again and filling in the gaps from memory. The two drift apart, and the gap between what you designed and what actually runs is where incidents live.
One source. Spec to running system.
The contract Capture writes already carries what Field needs. That is why this stage is an extension of the same pipeline, not a separate tool bolted on at the end.
Dev, staging, production. Same rules.
Every environment foots onto the same policy, read from the spec. Nobody re-types the deploy by hand and hopes it lines up, so the gap where that mistake lives is gone.
The deploy is in the contract already.
The things a person usually fills in by hand at deploy time are already settled in the spec. Field reads them. It does not reinvent them.
After the code. Before the world sees it.
Field takes the work Author built and puts it live under the spec's rules. It is the fourth stage of five, and it hands off to Track, which watches what you promised once it is running.
See the whole pipeline →