Decision: what it means
By Soumen Sarkar · October 2026
A decision is a call made in the work, recorded with its reason at the moment it is made, so it travels to everyone who touches the same ground.
A decision is a call that others should follow, written down where the work happens. It isn't made in a meeting or written up after the fact. The person or agent who makes it says what it is and why, in a sentence, and that sentence is the record.
The reason is what makes a decision useful later. "Use integer cents" is a rule someone might quietly break. "Use integer cents for money, because float rounding produced a one-cent refund error" is a rule people keep, and can revisit if the reason stops being true.
For example, an agent adding search to Docs · lib/search chooses Postgres full-text search over a separate search service. It records: "Search uses Postgres full-text search, because the corpus is small and it avoids running another service. Revisit past a million documents." The next agent that touches search sees it before it writes a line.
At agent pace, the risk is two agents making opposite calls in the same area without knowing. Recording decisions as they're made, and checking new ones against those that stand, catches that early.
The decision line works anywhere. In Campfyrr, log_decision checks a new decision against the standing ones; if it contradicts one, the newer call stands and the earlier one is marked replaced; decisions then appear in the area brief of every agent that touches that code.
The principle behind this: Decide by doing: Our decisions live in what we do, not in what we say.
Related: Superseded · Team record · Area brief
Questions that use it
Move as one at agent speed. Nobody waits. Add Campfyrr to Claude: it briefs every agent before it touches the code, keeps assumptions and handoffs where the team can see them, and never takes your code (what it keeps).