CLI Companion Standup

`ed companion standup`

Records a standup and, on request, checks what you said against what the record shows. The per instance result is nearly worthless. The aggregate is the point.

Usage:

ed companion standup [record] <file|-> [--verify] [--json] [--endpoint <url>]
ed companion standup report [--json] [--endpoint <url>]

ed companion standup record (also the bare default) reads a transcript from a file or stdin, stores it as an episode, and extracts the claims inside it: what you finished, what you committed to, how you rated yourself. --verify goes further and resolves those claims against the connectors, tracks the commitments and scores the calibration.

ed companion standup report is the aggregate: how many standups, what share of commitments landed, the median slip in days, and how often your account ran ahead of or behind the record.

--json shape for report: {standups, commitmentsResolved, metRate, medianSlipDays, overstated, understated, invisibleWork, dueSoon}.

Record JSON is {episodeId,occurredAt,verified,claims}. Each claim is {id,statement,claimType,testable,verdict,note}; verdict and note are null without verification or when the evidence is unclear. The file must be UTF-8 and nonempty after trimming. - reads stdin to end-of-file.

Read it as a calibration mirror rather than a lie detector. Almost all standup inflation is optimism, not deception, and optimism bias is the more actionable finding because it is fixable and it is not shameful. When work happened somewhere the connectors could not see, correct it with ed companion discrepancies override.