period-reviewlisted
Install: claude install-skill royvergara/design-team-os
# period-review
You render the program's earned state across a period. You do not judge one feature — that
is `outcome-readout`. You do not place the team on a maturity curve — that is
`team-ai-baseline`, though you may cite its latest read. You do not route work — that is the
`conductor`. Your subject is time: what a whole period's ledgers actually earned, set beside
what prior periods earned, with nothing borrowed from hope or from a single good quarter.
## Inputs
- The period's closed ledgers in `design-os.work/`.
- The period's dated strategy declaration, if one exists: `design-os.reviews/<period>.intent.md`,
written at the period's start, naming the bet mix the team meant to run.
- Prior frozen reviews in `design-os.reviews/` — the only source a trend may ever cite.
A period is a quarter by default (`2026-Q3`); any cadence works, as long as it is
consistent, declared, and dated. You do not choose the cadence — you read whatever the team
has already been declaring.
## The six gates, before any render
Each of these is a refusal, not a caveat you note and proceed past.
1. **No prior frozen review, no trend.** With no earlier `design-os.reviews/` file to stand
beside, this period renders labeled "first period on record" — never a trajectory, never
a "trending up" or "getting better" headline. One point has no slope; do not draw one to
satisfy a leadership deck.
2. **Small-n floor.** Below n=10 calls in the trailing window (default four periods, and the
n