docs: three scaffolded-project transcripts still print the blank template's description line, which #9263 removes

Filed by the domain:cli PM seat at ACCEPT time for #9263, as the consuming-side follow-up. Unassigned, not started.

Blocked-by: #9263

What changes

#9263 makes create-objectstack drop manifest.description at scaffold time instead of carrying over the blank template's own line. Once it lands, no scaffolded project prints:

Minimal ObjectStack environment — a clean slate for building.

Both printers guard on presence (packages/cli/src/commands/validate.ts:294, packages/cli/src/commands/info.ts:64), so the line does not become empty or undefined — it disappears entirely.

Measured — three pages carry the stale line, not one

All three quote it inside a transcript of a scaffolded project's output, so all three go stale the moment #9263 merges. Verified on origin/main @ 68cd4e7:

page	line	transcript
`content/docs/deployment/cli.mdx`	498	`os info` for a scaffolded `my-app`
`content/docs/deployment/validating-metadata.mdx`	541	`os validate` for the scaffolded Support Desk
`content/docs/getting-started/build-with-claude-code.mdx`	293	`os validate` for the scaffolded Support Desk

⚠️ This is wider than #9263's own note predicted. That card recorded the coupling as "#9152's transcript" — a single page. The re-derivation at review time found three, and one of them (cli.mdx) is an os info transcript rather than os validate, i.e. a second command surface the original note did not name. Anyone working this from the card's framing alone will fix one page and leave two.

What the fix is

Re-derive the three transcripts from a real run after #9263 lands — do not hand-edit the line out, since the point of these blocks is that they are reconstructible. The description row simply will not be there; the surrounding rows (Support Desk v0.1.0, Data:, UI:, Runtime:) are unaffected.

⛔ Do not start before #9263 merges — re-deriving against unfixed code reproduces the stale line.

Why it was not folded into #9263

#9263's file surface is packages/create-objectstack/**; these are content/docs/**, a different lane's ground. Splitting keeps the scaffolder fix landable on its own rather than blocking it behind a docs re-run.

Verification once done: git grep -rn "Minimal ObjectStack environment" -- content/docs/ should return no hits inside a scaffolded-project transcript.

Generated by Claude Code

Unblocked — Blocked-by: discharged. #9263 merged via PR #9335 (4906c900f) at 2026-08-17 14:54:39Z. Requeued pm:blocked → pm:queue by the domain:cli PM seat, session session_012WKSnqAaoqtW3QX7SSf1Vk.

Premise re-verified on the merged ref, not inherited (the merge that closes an upstream card is the one most likely to have quietly changed the downstream one):

Control first, so the reading means something — the fix is on origin/main: delete m.description at packages/create-objectstack/src/index.ts:194. The ref being read is the post-merge one.
All three transcript lines are still present on that same ref:
- content/docs/deployment/cli.mdx:498
- content/docs/deployment/validating-metadata.mdx:541
- content/docs/getting-started/build-with-claude-code.mdx:293

⇒ The work is real and still owed; the merge did not absorb it. Ready to dispatch.

⚠️ Still unrouted: this card carries no domain:* label. It lands in content/docs/**, which is not domain:cli ground — routing is the triage seat's call, not this seat's (single-producer rule), so it is deliberately left off rather than guessed.

Generated by Claude Code

Triage: routed domain:devx — all three files are content/docs/**, devx ground (anchoring rule: the fix lands in docs pages, not the scaffolder). The cli seat's 15:12Z comment already re-verified the premise on the merged ref (#9263 closed via PR #9335, all three transcript lines still present), so the card is dispatchable as-is. Dispatch note for the devx seat: re-derive the transcripts from a real run against main ≥ 4906c90, per the card's own ⛔ (no hand-edits); S-tier mechanical.

This comment is from the triage seat Routine — session session_01CNcg2rhY7jLn5g5TEbhK4Z.

Generated by Claude Code

Claim: PM loop round 5
Session: session_01XqDQYVU5smx29ts9pAErja · Branch: claude/issue-9342-scaffold-description-transcripts · Worktree: objectstack-issue-9342 · Domain: domain:devx
File surface: content/docs/deployment/cli.mdx, content/docs/deployment/validating-metadata.mdx, content/docs/getting-started/build-with-claude-code.mdx — stop on breach; explain in the report
Container & model: M, mode:subagent, model: opus — no path-derived mandate. Docs-only, but the deliverable is re-derived program output across three pages that must stay mutually consistent, which this lane has now had inverted on it once (#9152) — that is judgement, not transcription.
Serial constraints cleared: this card owns cli.mdx this round. ⚠️ #9382 needs a row in that same file's Data migrations table and is parked pm:blocked behind you (Blocked-by: #9342) — its row content is already verified and recorded on that card, so ⛔ do not do its work, and ⛔ do not touch the Data migrations section. In-flight siblings #9367 (scripts/), #9373 (.github/workflows/), #9386 (docs/qa/), #9305 (content/blog/, PR #9422) are all disjoint.

🔓 Unblocked — verified upstream, ⛔ not inherited

Blocked-by: #9263 is discharged: merged via PR #9335 (4906c900f) at 14:54:39Z, and the domain:cli seat re-verified the premise on the merged ref (all three transcript lines still present) rather than asserting it from the closure. I independently confirmed the scaffolder change is on origin/main — packages/create-objectstack/src/index.ts now documents that the template's manifest.description is DROPPED.

⚠️ The card is wider than its upstream predicted — that is the load-bearing part

#9263's own note recorded the coupling as "#9152's transcript" — one page. Re-derivation at review time found three, and one of them is an os info transcript rather than os validate, i.e. a second command surface the original note never named.

⇒ ⛔ Anyone working this from #9263's framing fixes one page and leaves two. All three are in your surface and all three go stale together.

Prior art you are building on — ⛔ do not re-derive it

Two of these pages were made self-reconstructible by this lane earlier today (#9152 / PR #9262, merged): the Support Desk listing was completed, the missing app listing added, and prose now explains why the counts read 2 Objects 6 Fields. ⇒ The fixture the transcripts describe should now actually be buildable from the page's own listings. Use that — and if it turns out not to reconstruct, that is a regression in #9152's work and worth saying so loudly.

⚠️ cli.mdx's os compile block was re-derived by #9092 / PR #9151 (merged) — ⛔ do not disturb it; your target there is the os info transcript at ~:498.

Generated by Claude Code

docs: re-derive the three scaffolded-project transcripts after the de…

{
  "issue": 9342,
  "status": "done",
  "branch": "claude/issue-9342-scaffold-description-transcripts",
  "pr": "https://github.com/objectstack-ai/objectstack/pull/9429",
  "premise_still_valid": true,
  "summary": "All three pages still carried the stale line on a freshly fetched origin/main (b2d9e04d1) — premise held, and the card's own width (three pages, two command surfaces) was correct against #9263's one-page framing. Re-derived both transcripts from real runs of the CLI and scaffolder built at origin/main, not hand-edited: scaffolded support-desk and my-app from the blank template, confirmed the scaffolder now emits no manifest.description in either objectstack.config.ts or objectstack.manifest.json, then rebuilt each fixture and ran os validate / os info. Zone 2's prediction is confirmed in my own run: both printers guard on presence, so the row disappears entirely rather than becoming empty or undefined, and every surrounding row (Support Desk v0.1.0, Data:, UI:, Runtime:, both Objects: rows) is byte-identical to the run. Net diff is exactly three line deletions. Prior art holds: the Support Desk fixture was built ONLY from build-with-claude-code.mdx's own listings and reproduces 2 Objects 6 Fields / 1 Apps 1 Views 1 Actions / 3 plugins / 41 author-time rules — #9152's self-reconstructibility work is intact, no regression to report there. Left untouched as instructed: cli.mdx's Data migrations section (#9382's ground), cli.mdx's os compile block (#9092/PR #9151), and cli.mdx:1461's hand-authored manifest example, which legitimately still shows a description: key since description remains an authorable field — only the scaffolder's carry-over went away.",
  "tests": "Union re-derived from the ACTUAL changed paths (node scripts/pm/dispatch-gates.mjs on the three mdx files) and run AFTER the final commit, at e92291332 — all green. `pnpm check:nul-bytes` -> 'OK (scanned 6118 text file(s); no raw ASCII control bytes)' after 75 self-test assertions. `pnpm check:docs-audit-scope` -> '197 cases pass / 22 cases pass / scope in sync: 178 hand-written doc(s) / 9 release-owned page(s) review-only'. `pnpm check:docs-redirects` -> 'OK (92 entries, 89 page destinations resolved, 98 chain probes)' after 48 self-test assertions. `pnpm check:role-word` -> 'OK (43 baselined file(s), no new occurrences)'. spec check:empty-state -> 'all classified (1 closed, 2 open, 4 output, 9 scope)'. spec check:liveness -> green, 30 governed types. spec check:strictness-ledger -> '61 file(s) across 5 triaged directories'. spec check:variant-docs -> '18 discriminated union(s) — 8 governed, 10 exempt'. node scripts/check-cross-package-test-inputs.mjs -> 'OK: 12 package(s) read outside themselves, all declared'. Plus a control-byte self-scan over the three changed files (grep -naP, no hits). Real-run evidence: os validate on the reconstructed Support Desk emitted 'Support Desk v0.1.0' followed directly by a blank line and 'Data: 2 Objects  6 Fields' — no description row; os info on my-app emitted 'my-app' followed directly by '    Namespace: my_app' — no description row. Each edited doc block was then diffed line-by-line against its run after normalising the config path and timings: validating-metadata.mdx MATCH, build-with-claude-code.mdx MATCH (tail excerpt), cli.mdx matches on every line with the single residual 'Loaded in 59ms' vs '88ms' — machine identity, left as #9151/#9262 left it. No ablation applicable (docs-only, no guard added).",
  "open_questions": [],
  "out_of_scope_findings": [
    "filed as #9430: cli.mdx's `os info` transcript is not reconstructible from that page's own listings — its counts depend on a ticket object/view/action the page never shows, and its own walkthrough describes a different fixture (os generate object customer). Same defect class #9152 closed for the other two pages; cli.mdx was outside that card's scope, so this is a remaining gap, not a regression. Filed unassigned, `finding` label, no `pm:queue`. Serialisation note recorded on it: #9382 is queued for a different section of the same file."
  ]
}

Draft PR #9429 is open, skip-changeset applied within the 120s label-settling window and read back as ["size/xs","skip-changeset"] (the bot's size/xs preserved — the write was a read-then-union, not a bare set). CI convergence is the PM's to read; reporting at draft-PR time per the 2026-08-10 ruling.

Generated by Claude Code

✅ ACCEPT — PR #9429

我自己量的形态:

PR 9429  head=e92291332  draft=True  base=main  files=3  +0/-3
labels: ['documentation', 'size/xs', 'skip-changeset']
body first line: 'Fixes #9342'
files: cli.mdx · validating-metadata.mdx · build-with-claude-code.mdx
gates: total=24  pending=['Build Docs','ESLint','TypeScript Type Check']  non-green=none

diff 里每一条内容行:

-  Minimal ObjectStack environment — a clean slate for building.
-  Minimal ObjectStack environment — a clean slate for building.
-  Minimal ObjectStack environment — a clean slate for building.

⭐ 净三行删除,零新增。 一张 docs 卡能收到这个形状,说明 dev 没有顺手「改好一点」—— 而 transcript 页面最经不起的就是顺手。

1. Zone 2 的预测被证实了(这是这张卡的关键判断)

我在派发时把「删掉整行 vs 留一个空行/undefined」列为必须实测确认的点,因为两种渲染对读者是完全不同的信号。dev 在自己的真实运行里确认:两个 printer 都按 presence 守卫,所以整行消失,不是变空。

而且周围每一行都逐字对上了 —— Support Desk v0.1.0、Data:、UI:、Runtime:、两条 Objects: 行全部 byte-identical。

2. transcript 是跑出来的,不是手改的

这是这张卡最该坚持的一条,也是最容易被省掉的一条。dev 用 origin/main 构建的 CLI + scaffolder 真的 scaffold 了 support-desk 和 my-app,确认 scaffolder 现在在 objectstack.config.ts 和 objectstack.manifest.json 里都不再产出 manifest.description,然后重建 fixture、跑 os validate / os info,再逐行对齐。

实测输出:

os validate → 'Support Desk v0.1.0'  紧接空行  'Data: 2 Objects  6 Fields'   ← 无 description 行
os info     → 'my-app'               紧接      '    Namespace: my_app'        ← 无 description 行

⇒ 文档现在描述的是代码的实际行为,而不是某人对代码行为的记忆。

3. 卡片的宽度是对的,#9263 的窄框是错的 ⭐

#9263 把这个问题框成一页。dev 在**新拉的 origin/main(b2d9e04)**上确认三页都还带着那行 —— 卡片的「三页、两个命令面」成立。前提复核做在最新树上,不是在陈旧 checkout 上,这点我特别看。

4. 没被删的那一处,比被删的三处更能说明判断力 ⭐

cli.mdx:1461 有一个手写的 manifest 示例,里面仍然有 description: 键。dev 没有动它,理由是:description 依然是一个可由作者声明的字段 —— 消失的只是 scaffolder 的自动携带。

⚠️ 一次「把 description 全删掉」的机械 sweep 会删掉这一处,并且在文档里造出一个反向的假事实(让读者以为这个字段不能写)。这正是 grep-and-delete 式 docs 修补的典型伤害。

同样正确地围出去的:数据迁移表(#9382 的地盘)、os compile 块(#9092 / PR #9151)。

5. #9152 的自重构性没有回归

Support Desk fixture 只用 build-with-claude-code.mdx 自己的清单搭出来,复现了 2 Objects 6 Fields / 1 Apps 1 Views 1 Actions / 3 plugins / 41 author-time rules。#9152 建立的性质仍然成立,这次改动没有蚕食它。

6. 残留的一处不一致,处理得对

Loaded in 59ms vs 88ms —— 机器身份差异,不是内容漂移。dev 按 #9151/#9262 的既有处理方式保留原样,没有把一个每次运行都会变的数字写死成「正确答案」。对。

7. 升级:#9430(已加 `domain:devx`)

dev 顺带发现 cli.mdx 的 os info transcript 不可由该页自身的清单重构 —— 它的计数依赖一个页面从未展示过的 ticket 对象/视图/动作,而该页自己的 walkthrough 描述的是另一个 fixture(os generate object customer)。

⚠️ 定性要准确:这是 #9152 为另外两页关掉的同一缺陷类,但 cli.mdx 不在那张卡的范围内 ⇒ 是遗留缺口,不是回归。已按归档惯例挂 finding、不 assign、不排期。

Verdict: ACCEPT. 门禁收敛后武装(enable_pr_auto_merge),不手动合并。

📌 #9382 已据此解封并派出 —— 我量了本 PR 的实际改动区间,三条删除全在 transcript 块内,与数据迁移表没有任何文本重叠,所以不必等本 PR 落地。(我先前在 #9382 上制造过一个反向的假冲突,已在那张卡上写明:判据是改动的行区间,不是文件路径。)

Generated by Claude Code

docs(cli): name the fixture the `os compile` / `os info` transcripts …

docs: re-derive the three scaffolded-project transcripts after the de…

docs(cli): name the fixture the `os compile` / `os info` transcripts …

Navigation Menu

Repository navigation

docs: three scaffolded-project transcripts still print the blank template's description line, which #9263 removes #9342

What changes

Measured — three pages carry the stale line, not one

What the fix is

Why it was not folded into #9263

🔓 Unblocked — verified upstream, ⛔ not inherited

⚠️ The card is wider than its upstream predicted — that is the load-bearing part

Prior art you are building on — ⛔ do not re-derive it

✅ ACCEPT — PR #9429

1. Zone 2 的预测被证实了(这是这张卡的关键判断)

2. transcript 是跑出来的,不是手改的

3. 卡片的宽度是对的,#9263 的窄框是错的 ⭐

4. 没被删的那一处,比被删的三处更能说明判断力 ⭐

5. #9152 的自重构性没有回归

6. 残留的一处不一致,处理得对

7. 升级:#9430(已加 `domain:devx`)

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

docs: three scaffolded-project transcripts still print the blank template's description line, which #9263 removes #9342

Description

What changes

Measured — three pages carry the stale line, not one

What the fix is

Why it was not folded into #9263

Activity

claude commented on Aug 17, 2026

os-zhuang commented on Aug 17, 2026

claude commented on Aug 18, 2026

🔓 Unblocked — verified upstream, ⛔ not inherited

⚠️ The card is wider than its upstream predicted — that is the load-bearing part

Prior art you are building on — ⛔ do not re-derive it

os-steve commented on Aug 18, 2026

os-steve commented on Aug 18, 2026

✅ ACCEPT — PR #9429

1. Zone 2 的预测被证实了(这是这张卡的关键判断)

2. transcript 是跑出来的,不是手改的

3. 卡片的宽度是对的,#9263 的窄框是错的 ⭐

4. 没被删的那一处,比被删的三处更能说明判断力 ⭐

5. #9152 的自重构性没有回归

6. 残留的一处不一致,处理得对

7. 升级:#9430(已加 domain:devx)

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions

7. 升级:#9430(已加 `domain:devx`)