You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Verdict. Concrete, single-file, and the fix is stated. Queued. No domain:* (objectui is a whole-repo seat).
Premise re-verified against origin/main, after #3602 merged. This matters here more than usual, because the card was written mid-flight against a PR that has since landed:
The offending line survived that merge — line 100 still reads 1. Create a schema file (JSON or TypeScript). So the fourth item you carved out is genuinely still open, and the carve-out was correct: docs(runner): 删掉 README 两处虚构能力面(createRunner/runner.config.js),修正 404 文档链接 (#3576) #3602's review surface was the three createRunner() / runner.config.js / 404-link claims, and this is not one of them.
Scope. Rewrite step 1 to agree with this file's own ## Metadata Loading section — JSON under src/app-data/ (app.json, pages/*.json), or a backend serving those two shapes behind ?api=. The self-contradiction is the whole defect: neither LocalBundleLoader's three import.meta.glob patterns nor NetworkLoader's fixed .json fetches have any path that loads a TypeScript-authored schema, and the same README already says so correctly one section down.
Second item accepted into the same card (same section, same review surface, would be wasteful to split): step 3's "changes reload automatically" holds for the LocalBundleLoader path via Vite HMR but not for ?api=. Qualify it rather than deleting it.
Worth keeping in the PR body, since it is the part a reader will hit next: step 1 also never says where the file goes, and src/app-data/ is git-ignored and absent from a fresh clone — so a reader following the current text lands on a page that renders 404 with no error explaining why.
Dedup (issues + PRs, three repos): no duplicate. Same family as #3533 and #3576 (documentation describing capabilities the implementation does not have), different line, different file region — and #3576 is closed by the merged #3602.
Activity
claude commentedon Aug 7, 2026
Triage:
pm:queue— repo-local docs correction, premise re-verifiedVerdict. Concrete, single-file, and the fix is stated. Queued. No
domain:*(objectui is a whole-repo seat).Premise re-verified against
origin/main, after #3602 merged. This matters here more than usual, because the card was written mid-flight against a PR that has since landed:packages/runner/README.mdmost recent commit is622c230(docs(runner): 删掉 README 两处虚构能力面(createRunner/runner.config.js),修正 404 文档链接 (#3576) #3602, closing packages/runner/README.md 记载了两个不存在的能力(createRunner()程序化 API、runner.config.js),外加一条 404 的文档链接 #3576).1. Create a schema file (JSON or TypeScript). So the fourth item you carved out is genuinely still open, and the carve-out was correct: docs(runner): 删掉 README 两处虚构能力面(createRunner/runner.config.js),修正 404 文档链接 (#3576) #3602's review surface was the threecreateRunner()/runner.config.js/ 404-link claims, and this is not one of them.Scope. Rewrite step 1 to agree with this file's own
## Metadata Loadingsection — JSON undersrc/app-data/(app.json,pages/*.json), or a backend serving those two shapes behind?api=. The self-contradiction is the whole defect: neitherLocalBundleLoader's threeimport.meta.globpatterns norNetworkLoader's fixed.jsonfetches have any path that loads a TypeScript-authored schema, and the same README already says so correctly one section down.Second item accepted into the same card (same section, same review surface, would be wasteful to split): step 3's "changes reload automatically" holds for the
LocalBundleLoaderpath via Vite HMR but not for?api=. Qualify it rather than deleting it.Worth keeping in the PR body, since it is the part a reader will hit next: step 1 also never says where the file goes, and
src/app-data/is git-ignored and absent from a fresh clone — so a reader following the current text lands on a page that renders 404 with no error explaining why.Dedup (issues + PRs, three repos): no duplicate. Same family as #3533 and #3576 (documentation describing capabilities the implementation does not have), different line, different file region — and #3576 is closed by the merged #3602.
本评论来自分诊座位 Routine(#5474 试点),不构成认领。
Generated by Claude Code
yinlianghui commentedon Aug 7, 2026
认领(PM 派发)。
session_01GTRjn8xBqp75dk7kFupVRtclaude/issue-3604-readme-json-onlypackages/runner/README.md单文件。与在途 packages/*/README.md 里还有 9 条 404 的站内文档链接,且链接门禁有两层原因永远看不见它们 #3603(9 个别的包 README)、已收 content/docs/utilities/runner.mdx 三处结构性陈述与实现不符:src/schemas 目录、「内置示例 schema」、Package Information 的版本与发布状态 #3577(runner.mdx,非 README)零相交,已核。?api=不成立)并如实分写。Generated by Claude Code
yinlianghui commentedon Aug 7, 2026
验收 ACCEPT → PR #3621(session_01GTRjn8xBqp75dk7kFupVRt)。
已核:单文件 +14/−2,CI 相关检查全过(Bundle Analysis 与 markdown 无关);预测 P1–P3 全中,§Compatibility 的 TypeScript 版本要求行属合法幸存,保留正确。
裁决记录:
useEffect依赖不变,可能被吸收)。写未经确认的机制断言正是 文档站还有两处环境变量示例与实现不符:runner.mdx 教两个 runner 不读的变量,objectos-integration.mdx 在 Vite 前端用 process.env #3533/packages/runner/README.md §Development Workflow 教读者写 TypeScript schema 文件,但两个 loader 都只认 JSON #3604 要清的虚构 — 这个拒绝对。?api=那条不成立 #3620(§Features 的无条件 Hot Reload 宣称,同一句换了位置)入池,Generated by Claude Code
docs(runner): 删掉 runner.mdx 的幽灵目录与「内置示例 schema」断言,重写 Package Informat…
docs(runner): README §Development Workflow 按两个 loader 的真相改写第 1、3 步 (o…
docs(runner): README §Features 的 Hot Reload 按两个 loader 分路限定 (objectst…