Added#
trazum verify <plan.json> --against <newer.jsonl>: did it work? Every
optimisation tool says what you would save; almost none says what you
did. This holds a saved plan to the log that came after it, with three
outcomes and never two: the change arrived, it did not arrive, or it
cannot be told — because the workload vanished, the fields the detection
needs stopped being recorded, or tokens cannot say which tier billed them
(the Batch API is invisible in token counts, and the verification says so
instead of assuming it happened). The third outcome is the honest one and
the one every other tool renders as the first.
Differences are attributed, not just stated: every plan action now records
its baseline (calls, dollars, tokens per call at plan time), and the
verification prints the world's movement beside each verdict — calls 3 → 6,
output/call 1,000 → 1,200 — so "the prediction was wrong" can be told from
"the traffic doubled". A plan priced under one catalogue and verified under
another is flagged (pricesChanged): the tool must not blame a team for a
saving that arithmetic revoked.
--gate makes promises checkable in CI: exit 1 when an action did not
produce what the plan promised, or became unverifiable because the team's
own log dropped the fields the detection needs — "not recorded" must not
read as "fixed". A workload that merely vanished fails nothing. --json
emits the verification document (contracted in docs/json-output.md);
--markdown-out writes the verdicts for a CI summary.
New core module verify.ts (verifyPlan, typed outcomes and reasons),
browser-safe: two documents in, one verdict out.