What an item is

An item is a record of intent. It carries the reasoning that produced it — the problem, the proposal, the costs weighed, how you will know it is done — it lives with the code it describes, and it is worth more after it closes than before, because it is then the answer to why is it like this.

Small tasks and architecture decisions can both carry useful reasoning. Write enough that a future reader can recover the choice without its original conversation. Closing the item makes it part of the project’s memory.

Cairn validates and changes that record. Git carries its history and review. Harrow is the terminal interface for watching the work and making decisions, using the same files and Cairn’s write commands.

Scheduling — statuses, milestones, what is ready to start — exists to make the record actionable, not the other way round.

What follows from it

This is not a slogan. It decides things.

• Items carry reasoning rather than titles, so a type’s template seeds headings — a feature rather than friction.
• cairn log was worth building, because a record whose history cannot be read is a worse record.
• The file format is specified for a horizon longer than this program’s.
• The answer to can filing be made faster by dropping the body is no.

What cairn should therefore refuse

A record of intent does not need comment threads, reactions, or a notification when somebody begins typing.

Those belong to the conversation, and the conversation is not the record — it is the thing the record exists to survive. That is the first place to point when somebody asks for a discussion feature.

The other boundary — what a single repository can and cannot do — is the promise.