Skip to content

Preserve saved case links when a research version cannot open - #8983

Merged
huangruiteng merged 2 commits into
mainfrom
codex/research-case-reading-20261118t0349
Nov 19, 2026
Merged

huangruiteng merged 2 commits into
mainfrom
codex/research-case-reading-20261118t0349

Conversation

@loopx-agent

@loopx-agent loopx-agent commented Nov 18, 2026 •

Copy link
Copy Markdown
Collaborator

Goal And Delivered Outcome

A saved case address may refer to a pack that has records but no authored study, or to a version that can no longer be read. The reader could replace the requested pack with another saved study and clear the chapter address. Preserve the requested version and give readers a way to recover in the same Goal.

  • Explicit version links retain the requested pack, with or without a chapter. If that version has no authored study, the version selector lets readers choose one.
  • An unreadable version can be retried at the same address. Browse saved studies opens the same Goal's available studies and restores keyboard focus. Browser Back returns to the original address.
  • A missing chapter stays within its saved version and directs the reader to the contents. A valid citation continues to open its exact original.
  • Closing the original with Escape returns focus to its citation. A delayed read of the same saved version preserves that focus and the chapter address.

Related to the saved citation navigation work in #8868. Target branch main.

Author Declaration

Criterion Disposition Surface Evidence
Saved versions retain their identity when reopened Implemented Case subview and existing version selector Explicit pack with and without a chapter, plus browser Back
Chapter and citation links use the saved version's source Implemented Existing Chapter and Citation readers Exact citation, foreign citation with no fetch, and missing chapter
Ordinary reading stays within the source and remains read-only Implemented Existing analysis reader and history view Retry and recovery in the same Goal, with no full-history reads or mutations

Scope And Continuation

The case view reuses the same URL state, analysis transport, source scope and version selector. Pack registration, evaluation, and execution retain their existing owners. The earlier version had an independent review. The updated commit awaits its own review and native desktop rollout.

Validation

  • Tested revision b38c02193f1c743bd6be62a92ece64d575cf15f1 on target revision 41ea75e89892ed88d0ae543e97e31aad0a181c47.
  • Run state finished. Input classes synthetic, authorized_private_read_only.
Check kind Result Public-safe evidence
static passed npm run build:chat type-checks the dashboard and builds and verifies the production Chat bundle.
regression_parity passed npm run smoke:research-evidence-version reaches the new recovery assertion and fails on the base bundle. The candidate passes the fixture with both explicit version links, exact retry, keyboard recovery, browser Back, missing anchors, and unchanged source data.
real_entrypoint passed npm run smoke:research-analysis-read passes. The packaged Chat entrypoint opens saved case addresses at 1200 px and 390 px.
real_backend passed The verified GitHub bundle for the earlier proposal opened the benchmark child, switched between saved versions and restored the original selection with Back. The actual source rejected a foreign version and the reader retained its address through retry and Back.
regression_parity passed A delayed read of the selected version causes the earlier GitHub bundle to fail the new focus assertion. The same synthetic fixture passes on the updated package with the original text and version isolation checks.
real_backend passed At 390 px, the updated package read all 11 lines of an existing source citation through the GET-only preview. After Escape, keyboard focus and the chapter address stayed in place through a controlled delay in the selected version read.
static passed Standard diff-scoped loopx canary premerge, exact-scope change-quality receipt, RFC pair check, public-boundary scan and git diff --check.

The earlier dashboard browser run failed in the Goal work-list expansion scenario at 184 of 200 visible items. The validation above covers the case reader in the updated package. Native desktop rollout awaits the normal release path.

Frontend / Visual Evidence

  • UI impact changed.
  • The captures show the version recovery layout with the same synthetic case fixture. The base is 3c03e91e67 and the packaged candidate is d1e0803882. I also inspected the packaged focus follow-up at 1200 and 390 px using the synthetic case fixture.

Before, desktop. The linked version has no study and the reader silently shows the initial version.

Base desktop showing a different version

After, desktop. The address retains its version and the page shows the selector alongside the empty state.

Candidate desktop retaining the linked version

After, 390 px. The missing version has separate retry and recovery actions. After recovery, keyboard focus lands on the version selector.

Missing version at 390 pixels Saved case and focused version selector after recovery at 390 pixels

The failure and empty states use the same case reader and its existing controls. The packaged preview is local to the validation environment.

Type Of Change

  • Bug fix
  • Documentation update
  • Test update

LoopX Area

  • Public docs or presentation surface (dashboard)

Technical Direction

S11, research evidence access in the LoopX roadmap.

Shared-authority RFC fixture impact

N/A. This changes the case reading view.

Boundary Checklist

  • Public-safe source, fixtures and screenshots only.
  • No new benchmark execution.
  • Published diff scoped to saved case navigation and recovery.
  • RFC bilingual diff check passed.
  • The first viewport and recovery states are embedded above.
  • The commit includes DCO sign-off.

@loopx-agent loopx-agent left a comment

Copy link
Copy Markdown
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Reviewer: model_agent · Gemini · Google (self_reported)

Approval conclusion (author-owned PR; GitHub blocks formal self-approval)

已针对 PR #8983 exact head d1e080388230b5b9a0a4a41f208596b122218531 完成独立代码审查、UI 首屏截图视觉核对与本地实测验证(审查结论:APPROVE,无 P0/P1/P2 阻断问题;本 PR 为作者自有 PR,依据规则以 COMMENTED 形式发布审查结论)。

动机 / Motivation

受影响调用方与用户:通过包含显式 research_pack 参数及可选 #case-section-... 锚点的持久化案例链接访问 Dashboard/Chat 研究视图的读者、在仅有研究记录而无案例解读或版本暂时不可读时需要使用版本选择器或返回同 Goal 既有解读的键盘与普通用户。

修改前(3c03e91e6736f129d78e7c895b9d4a871f132a7f,依据 docs/reference/trajectory-assessment.md#authored-case-studies 规格基线):当显式链接指向一个仅有研究记录而无 case_study 的 research_pack 时,ResearchCaseReading 未判断 a.location.packId === null 即自动跳转 a.select(firstStudy),并清空章节锚点;且 GoalResearchHistory 未在 requestedCase === true && analysis.location.packId !== null 时保留案例视图入口;当链接版本不可读时,AnalysisReadState 仅提供“重试读取”而没有在同一 Goal 内切换其他已保存解读的安全出口;ResearchCaseStudy 亦未区分缺失章节与缺失引文。修改后(d1e080388230b5b9a0a4a41f208596b122218531):显式版本链接始终保留目标 research_pack 与哈希锚点;无解读版本呈现结构化空状态并保留“解读版本”选择器供显式切换,浏览器 Back 返回原链接;不可读版本允许在原地址重试读取,并提供“查看已保存的解读”(Browse saved studies)在同一 Goal 内切换可用解读并恢复键盘焦点;缺失章节提示从目录选择而不跨版本改写来源,合法引用继续打开原文。

可观测结果:在 d1e080388230b5b9a0a4a41f208596b122218531 上实测:npm run build:chat(tsc --noEmit + vite build + 产物校验)全部通过;npm run check:css-custom-properties(43 份样式表、62 个 token、0 未定义引用)通过;npm run smoke:research-evidence-version 在桌面 1200×900 与窄屏 390×900 下通过,且对照试验将 HEAD 浏览器测试脚本针对基线 3c03e91e6736f129d78e7c895b9d4a871f132a7f 构建的 Chat 产物运行时,在新增“此版本没有已保存的案例解读。”断言(L254)处按预期超时失败;npm run smoke:research-analysis-read 与 npm run smoke:research-human-attention 通过;personal-workspace-browser-smoke 中 research-source-recovery、research-turn-costs 与 research-tool-context 场景均通过;loopx canary premerge --tier quick --from-git-diff --git-diff-base 3c03e91e6736f129d78e7c895b9d4a871f132a7f(ok=true, failures=0)与 loopx check --scan-path(5 文件,errors=0)均通过。PR 描述已直接内嵌公开安全的首屏前后对照截图(桌面 1200×900 与窄屏 390×900 恢复前/恢复后焦态)。

非目标边界:本 PR 复用既有 URL 状态、/api/chat/research-analysis 只读接口、来源范围与版本选择器,不改动轨迹包注册、评估、执行所有权或原生桌面打包发布流程。

改动思路 / Approach

架构与放置判断:直接复用既有 useResearchAnalysis、AnalysisReadState 与 ResearchCaseStudy 控件,不新增路由参数或服务端接口,仅在 ResearchCaseReading 守卫 packId === null、在 GoalResearchHistory 维护 caseViewAvailable、在 ResearchCaseStudy 统一 missingAnchor 状态并同步更新 3c03e91e6736f129d78e7c895b9d4a871f132a7f 基线的规格文档。

当前 PR 边界:仅限于研究案例解读子视图的链接保持、版本切换恢复、焦点管理与对应文档/冒烟(5 文件 +129 / -28,基线 3c03e91e6736f129d78e7c895b9d4a871f132a7f)。

用户场景 基线 3c03e91e6736f129d78e7c895b9d4a871f132a7f 当前 exact head d1e080388230b5b9a0a4a41f208596b122218531
显式链接到无解读的 research_pack 自动改写为 firstStudy 并丢失原版本地址 保留原 research_pack 与锚点,显示“此版本没有已保存的案例解读。”及版本选择器
显式链接到不可读的 research_pack 仅可重试,无同 Goal 出口 显示“无法读取链接指定的解读。”+ 原地址重试 + “查看已保存的解读”(恢复焦点至 select/h2)+ Back 返回
显式链接到缺失章节 #case-section-... 静默忽略 显示“这条链接在所选解读中找不到章节。可从目录选择。”而不读取研究历史

具体改动 / Changes

本次 diff(3c03e91e6736f129d78e7c895b9d4a871f132a7f..d1e080388230b5b9a0a4a41f208596b122218531)共修改 5 个文件(+129 / -28,无 RFC 文件变更),对照规格基线 docs/reference/trajectory-assessment.md#authored-case-studies(修订 3c03e91e6736f129d78e7c895b9d4a871f132a7f)的 3 项验收标准(saved_version_identity_retained、chapter_and_citation_use_saved_version、read_only_same_goal_recovery)逐一实现并验证:

  1. apps/presentation/dashboard/src/features/personal-workspace/goal-research-analysis.tsx(+28 / -10,对应 saved_version_identity_retained 与 read_only_same_goal_recovery):放宽 select: (id: string | null);扩展 AnalysisReadState 支持 onBrowseStudies 按钮(“查看已保存的解读” / “Browse saved studies”)及“无法读取链接指定的解读。”提示;ResearchCaseReading 新增 focusAfterBrowse useLayoutEffect 焦点恢复、限定 a.location.packId === null && !window.location.hash.startsWith(caseChapterHashPrefix) 才自动选首版、并在 !a.loading && !a.readError && p 时显示“此版本没有已保存的案例解读。”(role="status")。
  2. apps/presentation/dashboard/src/features/personal-workspace/goal-research-history.tsx(+6 / -5,对应 read_only_same_goal_recovery):引入 caseViewAvailable(含 requestedCase === true && analysis.location.packId !== null),确保显式指向无解读版本时视图切换器与 deferHistory 保持稳定而不意外退回并触发全量历史请求。
  3. apps/presentation/dashboard/src/features/personal-workspace/research-case-study.tsx(+16 / -11,对应 chapter_and_citation_use_saved_version):将 missingCitation 合并为 missingAnchor(字符串),为 <h2> 添加 tabIndex={-1} 以承载键盘焦点恢复,并按 missingAnchor.includes('~') 分别提示缺失原文引文与缺失章节目录引导。
  4. docs/reference/trajectory-assessment.md(+4 / -0):在 authored-case-studies 小节补充缺失章节、无解读版本及不可读链接的显式恢复行为说明。
  5. apps/presentation/dashboard/smoke/research-evidence-version-browser-smoke.mjs(+75 / -2):在桌面 1200×900 与窄屏 390×900 视口补充无解读版本保留(含/不含 hash)、版本选择与 Back 回退、不可读版本 409 原址重试与“查看已保存的解读”焦点检查、缺失章节提示与目录导航、以及 caseHistoryReads === [] / citation_reads / source_unchanged 不变量断言。

关键代码讲解

  1. ResearchCaseReading(goal-research-analysis.tsx:425-460):useEffect 严格校验 a.location.packId === null 防止覆盖显式版本 URL;browseStudies 仅在 a.readError && !p && a.location.packId !== null 时挂载,点击后置位 focusAfterBrowse.current = true 并调用 a.select(null),待新快照就绪且焦点落回 document.body 时在 useLayoutEffect 同步将焦点迁往 select、<h2 tabIndex={-1}> 或错误按钮。
  2. ResearchCaseStudy(research-case-study.tsx:81-145):对不匹配当前 pack 的章节/引文锚点,scoped 为 false,直接置 missingAnchor(hash) 并在状态栏分别引导从目录或依据入口重新选择,避免跨版本盲跳。

对主干的风险 / Risks

  • 无 P0/P1/P2 代码阻断问题:新增 UI 状态均严格约束于显式 research_pack 阅读路径,未改写分析范围、URL 历史记录规则以外的状态或触发额外后台请求。
  • CI dashboard-browser (3) 与残留验证说明:远端 CI 中 dashboard-browser (3)(及聚合 checks)在当前运行中报错,且 test-shard (1)-(6) 仍处于 PENDING。本地复跑 personal-workspace-browser-smoke 时,research-source-recovery、research-turn-costs、research-tool-context 均通过,而 capability-scope 与 research-worker-readback 返回预置的 403 (Forbidden);在基线 3c03e91e6736f129d78e7c895b9d4a871f132a7f 同等环境下实测 capability-scope 亦出现完全一致的 403 (Forbidden) 报错(与 PR #8983 改动无关)。正式合并与桌面打包仍需等待 CI 分片结果与独立维护者授权。

我的整体评价 / Verdict

改动简洁聚焦、边界清晰,准确解决了显式案例链接指向无解读版本或不可读版本时静默改写地址、丢失章节位置与缺失恢复出口的问题,键盘焦点管理、浏览器 Back 导航、390px 窄屏适配与只读/不读取研究历史不变量均经过严格验证;PR 正文亦提供了清晰的桌面端与 390px 端首屏前后截图对比。visual_review:inspected(原因:已直接检视 PR 内嵌的 4 组 1200×900 桌面端与 390×900 窄屏首屏截图并核对当前 head 构建的 Playwright smoke:research-evidence-version 实测结果:桌面端对比基线误跳 初版解读 的缺陷已变为保留 选择已保存的解读 选择器并展示清晰的 此版本没有已保存的案例解读。 状态;390px 窄屏缺版恢复视图将错误提示、重试、查看已保存的解读 及折叠结论垂直对齐且无横向溢出;恢复后 :focus-visible 焦点环准确落在全宽 解读版本(修订版解读)选择器上,正文与层级清晰可读。;未验证范围:远程 dashboard-browser (3) 全部非研究场景(本地 capability-scope 与 research-worker-readback 返回预置 403 (Forbidden),且 capability-scope 403 在基线 3c03e91e67 同等复现),以及待执行的全分片远程 CI 与桌面打包发布。)。结论为 APPROVE(作者自有 PR,依据规则以 COMMENTED 形式发布)。

English verdict: APPROVE exact head d1e0803 (author-owned COMMENTED approval review); keeps explicit research_pack links and anchors intact across records-only packs, unreadable versions (with same-address retry, Browse saved studies, and keyboard focus restoration), and missing chapters against docs/reference/trajectory-assessment.md#authored-case-studies (3c03e91e6736f129d78e7c895b9d4a871f132a7f), verified via npm run build:chat, check:css-custom-properties, smoke:research-evidence-version (with red-green base counterfactual failing at L254 on 3c03e91e6736f129d78e7c895b9d4a871f132a7f), smoke:research-analysis-read, smoke:research-human-attention, 3 research personal-workspace scenarios, canary premerge --tier quick (failures=0), and loopx check (errors=0); remote dashboard-browser (3) and pending test-shard CI checks remain gated on merge readiness (capability-scope 403 reproduces identically on base 3c03e91e6736f129d78e7c895b9d4a871f132a7f).

@loopx-agent loopx-agent left a comment

Copy link
Copy Markdown
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Reviewer: model_agent · gpt-6.2-sol · OpenAI (runtime_reported, max)

Approval conclusion (author-owned PR; GitHub blocks formal self-approval)

动机 / Motivation

这项修改服务于从 Benchmark 打开子 Goal、再用书签回查已保存案例的读者。读者打开记录中保存的链接时,旧版可能换到另一份解读。按 Esc 关闭原文后,同版的延迟响应还会把键盘焦点带回章节,读者需要重新定位引用。候选保留链接指定的版本和章节。原文关闭后焦点回到引用,收到同版数据时仍停在引用上。读不到版本时可在原地址重试,或显式进入同一 Goal 的解读目录。

正式 App 的安装与采用、真实长篇研究的可读性和实验因果验收各有单独的验证路径。隔离打包版已核验阅读和恢复。正式 App 内 ANN、Codec 的重新打开与实际采用仍待验收。

改动思路 / Approach

恢复是既有研究页的阅读逻辑,共用已注册版本、引用读取和浏览器历史控制。这次把章节定位绑定到真实导航,把显式恢复交还给版本选择器。本次交付覆盖子 Goal 的保存链接、失效恢复、引用往返和普通研究历程入口。 只在恢复焦点时加判断,仍会让失效链接误跳版本。现有控制器持有 Goal、来源、版本和原文请求;只读分支借用同一套选择、重试和前进后退机制。错误时保留原地址,使读者能核对请求后再决定是否换版本。

具体改动 / Changes

评审提交 b38c02193f1c743bd6be62a92ece64d575cf15f1。阅读约定固定在 41ea75e89892ed88d0ae543e97e31aad0a181c47,核对了四项行为。 Authored case studies

  • original-child-evidence。父页进入子 Goal 后仍可返回父页。指定的版本和章节由子 Goal 的同一来源读取。浏览器前进、后退均恢复合适的阅读入口。
  • focus-restore。Esc 把焦点交还原引用。同一版本回读晚到时,焦点和章节地址保持原位。
  • explicit-recovery。链接读取失败后可在原地址重试。同一 Goal 的保存解读由读者主动打开。缺章节显示目录,错来源的引用不请求原文。
  • saved-original-language。切换界面语言时保留保存的原文,并提示原文语言。

关键代码讲解

  • ResearchCaseReading 让默认推荐只作用于未指定版本的入口。真实读取错误可原地重试,明确选择浏览解读后,焦点落到可用的选择控件或标题。 research reader
  • GoalResearchHistory 依据现有 URL 与版本目录决定显示案例或历程。保存的版本仍是阅读目标,普通无解读的父页继续呈现原历程。 research history
  • ResearchCaseStudy 从当前保存的正文查找引用。仅在导航身份变化时恢复章节,防止换成同版的新数据对象时抢走链接焦点。 case study

对主干的风险 / Risks

同版延迟响应是本次最容易漏测的反例。独立打包对照使用同一个隔离后端,前一提交确实把焦点移到了章节,当前提交仍留在引用。旧版本的迟到结果也无法在切版后打开原文。原文读取注入 503 后,重试复用了相同的 Goal、版本、语言和行范围。错源引用没有发出原文请求,普通父页历程仍可读取。当前头的本地标准预合并、RFC 双语检查、研究原文读取回归和变更范围核验通过。

较旧的工作列表测试有 184/200 的诊断记录,具体成因仍未知。这条记录保留为历史诊断,本 Goal 的评审策略以当前提交的必要本地验证为准。 Earlier dashboard run

语义与 CI 对齐

改动复用现有研究读取约定。浏览、切版和引用重试均为只读,来源、权限、原始分析与执行入口仍由原有能力控制。实际隔离回放没有触发写入。

我的整体评价 / Verdict

阅读长程案例时,读者可沿同一历史版本反复回查证据,失效时保留地址并可继续。1200 像素的章节正文、目录与未测量样本同时可读。390 像素下,父 Goal 返回入口、版本选择和结论在同一页,展开引用后可看到返回的键盘焦点。正式安装版的长篇真实案例和读者实际采用尚未验证。 三个既有组件承接同一阅读流程,测试覆盖重现和恢复。这个范围可以独立评审,原生采用和科学研究验收继续保留。

English verdict: APPROVE b38c021. The packaged reader retains exact case links and keyboard focus after delayed reads. Same-source recovery, ordinary history and required local validation passed.

@huangruiteng
huangruiteng force-pushed the codex/research-case-reading-20261118t0349 branch from b38c021 to 25fe812 Compare November 19, 2026 06:59
Signed-off-by: LoopX Agent <337587101+loopx-agent@users.noreply.github.com>
Signed-off-by: LoopX Agent <337587101+loopx-agent@users.noreply.github.com>
@huangruiteng
huangruiteng force-pushed the codex/research-case-reading-20261118t0349 branch from 25fe812 to e7ae172 Compare November 19, 2026 07:38

@huangruiteng huangruiteng left a comment

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Reviewer: model_agent · GPT-5 · OpenAI · self_reported

Approval conclusion

动机 / Motivation

研究案例的保存链接应指向一个确定的证据版本,而不是给出“最接近可用”的提示。旧视图在目标 pack 只有研究记录或暂时不可读时,可能自动切到另一份解读,也可能在同版延迟回读到达后把键盘焦点从引用移走。候选保留显式版本和章节,提供原地址重试、同 Goal 浏览和明确空状态。

affected_caller_or_operator 是 Research readers and benchmark case reviewers;before_after_scenario 是显式版本链接保留原 pack;无 study 时显示选择器,不可读时可在原地址重试或显式浏览其他解读。observable_outcome 是原版本、章节和焦点保持,Back 回到原地址,错版本不会触发原文读取或写操作;non_goals 是不修改研究包注册、评估执行、URL 协议、后端分析请求或桌面打包流程。

改动思路 / Approach

现有 URL 状态、analysis controller、version selector 和 citation reader 已拥有完整职责;修复只需收紧这些组件的行为。显式 research_pack 决定阅读身份,加载/失败只影响状态展示;章节和引用恢复只在真实导航身份变化时发生,不再因新的 study object 替换而重跑。只修复已保存研究案例链接的版本保持、显式恢复、章节提示和延迟读取焦点,不改变研究包注册或后端执行。

具体改动 / Changes

精确 head e7ae172ffb61496f19d46529d83a5c73c1db3f5f,diff base 1e3c88bac9,5 个文件 +179/-30。规范引用 docs/reference/trajectory-assessment.md @ 1e3c88bac907f613cd4b0b1b4be27bc98e50b927;saved-version-identity、citation-focus 和 read-only-recovery 均实现。

关键代码讲解

  • ResearchCaseReading 只在没有显式 pack 时自动选默认 study,并为不可读版本提供同 Goal 的显式浏览与焦点恢复。
  • ResearchCaseStudy 用最新 study ref 读取引用,但 effect 只由 URL 导航身份驱动;延迟同版数据不会抢走关闭原文后的焦点。
  • GoalResearchHistory 的 caseViewAvailable 让显式版本保持案例视图,同时允许普通未绑定入口继续回退研究历程。

对主干的风险 / Risks

最强风险是 UI 修复在慢响应时仍抢焦点,或恢复按钮触发额外 history/source 读取。当前 packaged Chat smoke 控制延迟响应,检查 activeElement、版本地址、章节地址、citation reads 与 mutation count;records-only、unavailable、missing chapter 和 Back 均有反例。PASS:npm run build:chat 通过,npm run smoke:research-evidence-version 通过,quick premerge 通过。正式桌面壳和超长真实研究仍待单独验收。

语义与 CI 对齐

案例链接是只读证据地址:读取不会改变研究包、评估或执行状态。显式版本选择与 browser Back 都由现有 URL/history 机制持有,UI 不新增权限。

我的整体评价 / Verdict

批准当前精确提交。长期上,研究者能沿稳定地址反复返回同一版本及其引用;用户体验上,错误恢复是可见、键盘可达且有明确来源的空状态,不再静默改版。实现复用现有阅读组件和 URL 状态,新增测试主要锁定真实浏览器时序。

复用 PR 中已有的桌面与 390px 前后对照截图,并核对当前 head 是同一 UI patch 在最新 main 上的干净 rebase;当前浏览器 smoke 再次覆盖相同状态。unverified_scope:Native desktop shell and long-running production research data.

English verdict: APPROVE at e7ae172. Explicit saved study versions keep their pack, chapter and focus; records-only/unreadable cases recover in place, and the packaged Chat build plus browser smoke pass.

@huangruiteng
huangruiteng merged commit 14dbfcb into main Nov 19, 2026
10 of 16 checks passed
@huangruiteng
huangruiteng deleted the codex/research-case-reading-20261118t0349 branch November 19, 2026 07:47
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants