CLI Companion Why

`ed companion why`

Prints the whole chain behind anything the companion believes: the evidence episodes, what argues against it, the prompt version that produced it, how its confidence moved, and how it was checked against the record. When the system says something about you that feels wrong, this is how you find out whether it is wrong or you are.

Usage:

ed companion why <id> [--json] [--endpoint <url>]

The id can name a belief, a hypothesis or a claim, and the output follows what it found:

a belief prints confidence, stability, corroboration, the episodes it rests on and the episodes that argue against it
a hypothesis prints its mechanism, alternatives and recorded posterior revisions; the current CLI does not include its predictions
a claim prints when you said it and every verdict the record returned

No claim is allowed to bottom out at "the system concluded this". Every chain runs down to something you actually said or something a connector actually saw.

--json shape: {kind, id, statement, status, confidence, stability, corroboration, promptVersion, mechanism, prior, posterior, alternatives, evidence, counterEvidence, episode, revisions, verdicts}, with the fields that do not apply left null. The top-level confidence, stability, prior and posterior values are formatted as four-decimal strings when present; revision posteriors remain JSON numbers. Fields that do not apply are null for scalar values and empty arrays for list values. The API has additional belief links and dates, hypothesis test and prediction data, and claim metadata that the current CLI client does not expose.