CLI Companion Correct

`ed companion correct`

Retires a belief that is wrong, or rewrites it in your own words. This is non-negotiable rather than a convenience: if the system forms a wrong belief about you and you cannot correct it, every later answer inherits the error and you stop trusting the whole thing.

Usage:

ed companion correct <id> --retire [--json] [--endpoint <url>]
ed companion correct <id> --edit "<what it should say>" [--json] [--endpoint <url>]

--retire marks it retired. --edit writes a new belief carrying the same evidence, marked as yours rather than the extractor's, and supersedes the old one with a pointer back to it. Neither deletes anything: the append-only rule holds all the way up, so ed companion why can still show what was believed and when it stopped being believed.

Find the id with ed companion beliefs, and read the evidence behind it with ed companion why before you decide. Sometimes the belief is right and the surprise is the point.

At least one of --retire or --edit is required. An edit must contain more than five non-whitespace characters because the backend embeds the replacement. If both flags are passed with a valid edit, the edit takes precedence and the old belief is superseded rather than retired. JSON output is {id,status,statement}; for an edit, id is the new replacement belief.