feat(content): show what the content is, and demote the three-value key (split Step 2) - #128
Merged
Merged
Conversation
wu908
force-pushed
the
feat/taxonomy-split-step2-display
branch
from
September 16, 2026 09:55
c0c4f2f to
f1e0d67
Compare
…ey (split Step 2) Step 1 (8985c01) started recording an AI-named form on the deliverable and carried it onto the project at pickup, but nothing read it. The interface still opened every piece by asserting one of three behaviours: the shelf card said 「分享」, the list row said 「分享内容」, the workspace header said 「这是一条分享内容：…」, and the confirm action asked 「确认这是一条“分享”内容吗？」. That is what the creator pushed back on — three values read as three kinds of content, so the machine's routing key looked like a limit on what they were allowed to make. This step changes the display layer, not the data: - shelf card, content list, workspace outline: the AI-named form leads; the three-value label only remains as a fallback for drafts and projects written before the field existed (the two are shown side by side on the local shelf, and the list shows both cases). - pickup panel: 「AI 把它读成「水彩作品展示」」 above the confirmation, plus one line saying what the three values actually decide — which questions come next, how the piece is structured, which signals are watched. They do not constrain what it can be. - confirm panel: the heading is the open question 「这篇要读者拿走什么？」, the reader change is asked first, and the behaviour select is demoted below it. - workspace header: states the identity line only (「这篇是「水彩作品展示」。」); the reader change moves to 「这篇内容要完成什么」, which now prefers the project's own sentence over the per-intent generic one, so the two blocks stop saying the same thing twice. - the audience-change helper now says where that sentence came from (the project's own direction vs. the class's generic one) instead of always calling it a generic direction. Also splits task from question on that step: the action card now reads 「先把这篇的目的定下来」 while the panel asks 「这篇要读者拿走什么？」. They were briefly the same sentence, which put three identical headings on one screen — the E2E strict-mode locator caught it (and it read badly, too). The starter-flow spec asserted the old heading and was updated with it. Two defects found while verifying on the real stack: - the new deliverable view key was added without the response contract (DeliverableView is a StrictModel), so discard/pickup answered 500 with 「Extra inputs are not permitted」. The API suite caught it. The field is declared without a length cap on purpose: capping on the output side would turn one slightly long AI name into a 500. - reusing a pending action returns its stored title, so users would keep reading 「确认这是一条“解决”内容吗？」 after this change — and the row on the local server still named an intent (解决) the project no longer had (分享). A title is a derived label, not user input, so reusing an action now realigns title/reason with the current spec. Verified by reverting the fix: the new test fails without it. Verification: backend 507 passed (ruff clean), frontend vitest all passed with the coverage gate, tsc clean, eslint clean. Real-stack pass at 1280 and 390 with no overflow (frontend/scripts/taxonomy-step2-shots.mjs; screenshots in browser-screenshots/taxonomy-step2). The pickup in that pass was a real one: content_form travelled from the deliverable to the project, and the workspace header rendered it. Step 3 (rubric renaming) and Step 4 (widening the CHECK constraints, only if a fourth behaviour appears) are untouched.
wu908
force-pushed
the
feat/taxonomy-split-step2-display
branch
from
September 16, 2026 10:15
f1e0d67 to
14b5684
Compare
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Step 2 of the taxonomy split (#126). Step 1 (#127) recorded an AI-named form; this step is the display layer that reads it.
The complaint this answers: 「内容意图为什么要设定为三个枚举型，这样不会导致内容的局限吗？」 Three values were opening every piece — 产出架卡片「分享」、列表行「分享内容」、工作台表头「这是一条分享内容：…」、确认动作「确认这是一条“分享”内容吗？」. They were the machine's routing key, but they read as three kinds of content.
What changed
顺带修掉两个真问题（都在真实栈上验证时暴露）
DeliverableView是 StrictModel，extra=forbid）→ discard/pickup 全部 500，报Extra inputs are not permitted；输出侧刻意不设长度上限验证：后端 507 passed / ruff 干净；前端 vitest 全绿含覆盖率门禁；tsc、eslint 干净；真实栈 1280 与 390 走查（
frontend/scripts/taxonomy-step2-shots.mjs，截图在browser-screenshots/taxonomy-step2/），无横向溢出。那次走查里的认领是真实的：content_form从产出带到项目，工作台表头渲染的就是它。未做：Step 3（rubric 改名）、Step 4（只在出现第四种行为时才扩 CHECK）。
按约定，这一步先看截图再合并，PR 保持打开状态。