Migrate the in-repo schema corpus (~424 files) to carry the canonical `$schema` marker — the producer-side half of #5127's Option B

[os-support-ai]

Filed at dispatch time by the repo:objectui execution seat (PM round 17, session session_01RV6yuVCxymHYE16PL9vQkE), because the maintainer ruling on #5127 says so in as many words. Filed unassigned, not claiming.

Blocked-by: #5127

The ruling that creates this card

Maintainer, 2026-08-19, verbatim 「全部接受」, recorded on #5127:

B: complete direction C producer-side — decide the canonical ObjectUI $schema URL and migrate the ~424-file in-repo corpus so real schema files carry the positive marker. The URL proposal goes in the implementing PR for the maintainer to confirm at merge (it is a permanent public URL); the bulk migration is its own card, filed at dispatch time.

#5127 delivers the marker predicate, the URL proposal and the skipped-file count. This card delivers the corpus that makes the marker match. Until it lands, the repo sits in the interim the ruling explicitly permits and explicitly requires to be recorded as a debt, not as a closed gap.

The debt this card pays off

Measured during #5127's first dispatch, on origin/main 118419214 (re-measure before starting — the corpus and the numbers move):

marker set	real schema files retained (of 424)	exit surface
children alone	109	315 (74.3%)
children + body	118	306 (72.2%)
+ props, items	139	285 (67.2%)
full node vocabulary	150	274 (64.6%)
+ id	156	268 (63.2%)

Two structural causes, neither fixable by moving the structural-key line:

1. About two thirds of the corpus is single-node leaf schemas — {type: button, label: Send Email, icon: mail, variant: default}. A leaf has no structural key by construction.
2. The $schema arm is empty today — zero in-repo files point $schema at ObjectUI; the only judged file carrying $schema at all points at json-schema.org.

The marker is not failing on precision — every variant correctly retains zero foreign files. It fails on recall, and this card is the recall fix.

Scope when it unblocks

• Add the canonical $schema line to the real in-repo schema files — principally examples/schema-catalog/src/schemas/** (423 files, all imported as schemas by examples/schema-catalog/src/index.ts) plus examples/hello-world/schema.json.
• ⛔ Use the URL exactly as confirmed by the maintainer on objectui check 把每个 JSON 文件的根 type 都当成组件键判定,于是在任何 Node 工程里都对 package.json 的 "type": "module" 报未知类型 #5127's PR. Do not invent, shorten or "improve" it — a permanent public URL is a maintainer decision, and a second spelling in the corpus is worse than no marker at all.
• Verify by re-running objectui check and showing the skipped-file count (objectui check 把每个 JSON 文件的根 type 都当成组件键判定,于是在任何 Node 工程里都对 package.json 的 "type": "module" 报未知类型 #5127's D arm) fall to its floor. That number is the acceptance criterion — the migration is done when the count stops naming real schema files.
• Side benefit already on record: packages/vscode-extension/schemas/objectui-schema.json is a built-but-unused $schema consumer, so the migration should also buy editor autocomplete and validation. Check whether it does, and say so.

Why it is not a rider on #5127

Two different kinds of change: #5127 is a predicate plus a public-identifier proposal, reviewable in one screen; this is a ~424-file mechanical sweep whose diff would bury it. Keeping them apart is also what lets the maintainer confirm the URL before it is stamped into four hundred files.