Skip to main content
A standard is a protocol that fires on a file rather than on a prompt. Where a domain matches what you typed, a standard matches what you are about to edit, and it injects on mutating operations rather than on every turn.
base std is the short alias.

​
Where they come from

Why it works. Standards are authored in a protocols.md source and synced into standards.toml plus the graph. That is deliberately different from rules, which you add one at a time from the command line: a standard is usually a document that already exists, and the sync is what turns it into something that fires. The gotcha. There is no base standards add. If you want a new standard, write it into the protocols source and sync. Reaching for add and finding nothing is the most common first stumble here.

​
Proving one will fire

Why it works. This is a dry run of the matcher against a real file. It prints the scores and what would inject, and it writes nothing and sends nothing. --content simulates the payload of an actual edit, which matters because a standard can match on what you are writing as well as on which file you are writing it to. The gotcha. Test before you rely on one. A standard that does not fire is indistinguishable from a standard that fired and had nothing to say, and this command is the only thing that separates them.

​
Seeing what is loaded

Every standard with its trigger and annotation counts. A standard with zero triggers will never fire, which is worth noticing before you spend an afternoon wondering why. base standards get <id> shows one in full.

​
Where standards fit

Three mechanisms, three different moments. Standards are the one that reaches you while you are changing something rather than while you are asking about it.

​
Next

Rules and domains

The keyword-triggered half of the same idea.

Context reference

Every base standards verb and flag.