Share IMAGES_PLACEHOLDERS from @weaverse/schema

[hta218]

Summary

IMAGES_PLACEHOLDERS was defined only in @weaverse/hydrogen, so @weaverse/next storefronts could not import it. This PR moves the constant into @weaverse/schema. Both adapters already export * from '@weaverse/schema', so each now exposes it with no extra re-export. The values and their docs are unchanged, and import { IMAGES_PLACEHOLDERS } from '@weaverse/hydrogen' still works.

Changes

• Move IMAGES_PLACEHOLDERS from packages/hydrogen/src/index.ts to packages/schema/src/images-placeholders.ts and export it from @weaverse/schema
• Pin @weaverse/schema to 0.15.0 (was 0.14.0) in @weaverse/hydrogen and @weaverse/next, and refresh pnpm-lock.yaml
• Regenerate api-reports/: the constant moves from hydrogen.api.md to schema.api.md, and the runtime export report adds it to @weaverse/schema and @weaverse/next
• Add spec .specs/2026-09-15--images-placeholders-in-schema/

Why the pins move to 0.15.0

Internal @weaverse/* dependencies resolve from npm. scripts/check-packed-packages.mjs only uses the packed schema tarball when a package's pin equals the workspace schema version. With the old 0.14.0 pins, packed hydrogen and next would install a registry schema without the constant, and their runtime exports would no longer match the report. 0.15.0 is the current workspace version and is already on npm, so pnpm install --frozen-lockfile keeps working.

Release order (required)

A published @weaverse/hydrogen pinned to a schema without the constant would drop IMAGES_PLACEHOLDERS from its public API. Publish in this order:

1. @weaverse/schema 0.15.0 -> 0.16.0
2. Re-pin @weaverse/schema to 0.16.0 in packages/hydrogen/package.json and packages/next/package.json before building them
3. Fixed group (core, react, hydrogen), then the next @weaverse/next prerelease
4. pnpm install once the new versions are on npm, then commit the lockfile

Verification

biome ci ., pnpm run typecheck, pnpm run test and pnpm run package:check pass locally.

[claude]

Claude finished @hta218's task in 1m 40s —— View job

---

Claude finished reviewing this PR

Todo list

• Gather context (repo checkout, PR diff, CLAUDE.md conventions)
• Run code review on PR Share IMAGES_PLACEHOLDERS from @weaverse/schema #527
• Post review findings

Review

This is a clean, well-scoped move. I verified the diff against origin/main matches the PR description exactly (11 files, +202/-133).

Verified correct:

• IMAGES_PLACEHOLDERS (packages/schema/src/images-placeholders.ts) is a byte-for-byte relocation of the constant and its TSDoc from packages/hydrogen/src/index.ts — values and comments unchanged.
• packages/schema/src/index.ts:388 re-exports it with the .js extension, consistent with the existing ./validation.js / ./page-seo.js convention in that file.
• Since packages/hydrogen/src/index.ts and packages/next already do export * from '@weaverse/schema', both adapters keep exposing the constant with zero extra code — confirmed in api-reports/runtime-exports.api.md: the @weaverse/hydrogen export list still contains IMAGES_PLACEHOLDERS unchanged (flows through the wildcard re-export), while @weaverse/schema and @weaverse/next gained the new entry, matching the stated rationale.
• The @weaverse/schema pin bump 0.14.0 → 0.15.0 in packages/hydrogen/package.json and packages/next/package.json is justified: scripts/check-packed-packages.mjs's acceptsPackedVersion(range, packedVersion) (line ~90) does a strict range === packedVersion equality check, so a stale 0.14.0 pin would have caused the packed-package check to install the real npm-published schema (without the constant) instead of the packed workspace tarball — exactly as described.
• pnpm-lock.yaml and api-reports/schema.api.md / runtime-exports.api.md regenerations are consistent with the source change (no stray diffs).
• Spec at .specs/2026-09-15--images-placeholders-in-schema/ follows the required template (Status/Owner/Created/Last Updated table, self-contained Initiating Requirement, plan.md under 500 lines with a "Files touched" list, release-order section called out explicitly).

Minor, non-blocking nits:

• The spec's Status field is in-progress even though the PR appears feature-complete and ready for review/merge — consider updating to completed (or the appropriate status) once merged, per the SDD convention.
• The PR body's "Release order" section is a good callout, but since it's operationally important (skipping it would silently drop IMAGES_PLACEHOLDERS from a published @weaverse/hydrogen), it may be worth also noting it in work-logs.md or as a comment near the pinned version fields so it isn't lost outside the PR description.

No correctness, security, or style issues found. I was unable to execute pnpm/biome/turbo commands directly in this sandboxed environment (commands required interactive approval that isn't available here), so I relied on static diff review plus the PR author's reported local verification (biome ci ., typecheck, test, package:check all passing).
· Branch: update/move-images-placeholders-to-schema