[Hylouis233]

概要

将上游 pingdotgg/t3code main 同步到 fork main（前一次同步为 #31，同步到 680152e27b）。

• 上游 main：0bdf33a34dc520d3ab1364cd116686b3492184bd（refactor(web): the add environment dialog owns its form state (refactor(web): the add environment dialog owns its form state pingdotgg/t3code#22686)）
• fork 基线 origin/main：b001e934078f9605cfd0198c19069bd7f6a64dea
• merge base：680152e27b2a961686389ca69a1557cbb2c55f08
• 本次带入的上游提交（4 个）：
  • 9179178fdf docs(agents): proof for a bug fix depends on the bug (docs(agents): proof for a bug fix depends on the bug pingdotgg/t3code#23008)
  • fa42465ff6 fix(server): removing a worktree on Windows no longer deletes what its junctions point to (fix(server): removing a worktree on Windows no longer deletes what its junctions point to pingdotgg/t3code#21099)
  • 5a0d1594e8 perf(web): a streaming thread re-renders only its own sidebar row (perf(web): a streaming thread re-renders only its own sidebar row pingdotgg/t3code#22730)
  • 0bdf33a34d refactor(web): the add environment dialog owns its form state (refactor(web): the add environment dialog owns its form state pingdotgg/t3code#22686)

提交

1. 85189f76d5 Merge upstream pingdotgg/t3code main 0bdf33a into fork main（git merge，未 rebase）
2. d8e659a980 fix(ci): drop cursor-hygiene-webhook workflow to match upstream

冲突及处理

无冲突。apps/server/src/vcs/GitVcsDriverCore.test.ts 由 git 自动合并（fork 与上游都改过该文件），已单独跑过其测试（见下）。

• 合并后 git diff origin/main HEAD 涉及的文件集合与上游 680152e27b..0bdf33a34d 的文件集合完全一致（14 个文件），fork 独有改动未受影响。
• ci.yml（setup-vp、PR Gate 结构）、release.yml（仓库守卫）、model-manifest.json（orchestrationProtocolVersion：opencode/1 <0.0.46 = 1，opencode >=0.0.46 = 2，pi/2 = 2）本次均未改动。

删除文件检查

git diff --diff-filter=D origin/main HEAD：

• 合并提交本身：无删除文件。
• 第二个提交：.github/workflows/cursor-hygiene-webhook.yml。该文件来自上游（b5febb1b47 添加），上游已在 3e250a8387 删除（3e250a8387 是 merge base 的祖先，所以 merge base 中已无此文件）；fork 曾在 e37d4dd445 恢复了一份带守卫的版本。它不是 fork 独有文件。

测试改动与 workflow 删除

上游在 3e250a8387 删除了 .github/workflows/cursor-hygiene-webhook.yml。为与上游保持一致，本 PR 删除 fork 中的该文件，并调整 .github/scripts/opent3code-release.test.cjs 中的测试 “upstream APK publication and Cursor forwarding cannot run in the fork”：

• 文件不存在时，Cursor 转发不可能运行，测试视为通过；
• 文件存在时，仍要求 forward job 带有 if: github.repository == 'pingdotgg/t3code' 守卫；
• APK 发布检查（mobile-android-apk.yml / mirror）不变，文件缺失或缺少守卫依然失败。

-  for (const [workflow, job] of [
-    ["mobile-android-apk.yml", "mirror"],
-    ["cursor-hygiene-webhook.yml", "forward"],
+  for (const [workflow, job, optional] of [
+    ["mobile-android-apk.yml", "mirror", false],
+    // Upstream deleted this workflow (3e250a8387); when it is absent, forwarding cannot run.
+    ["cursor-hygiene-webhook.yml", "forward", true],
   ]) {
-    const source = fs.readFileSync(path.join(__dirname, "../workflows", workflow), "utf8");
+    const workflowPath = path.join(__dirname, "../workflows", workflow);
+    if (optional && !fs.existsSync(workflowPath)) continue;
+    const source = fs.readFileSync(workflowPath, "utf8");

同时更新 docs/operations/pr-actions.md 中关于该 webhook 的描述（仓库内唯一的其他引用；knip 配置中无引用）。

验证（仅运行该测试）：

场景	结果
删除 workflow，测试未修改	失败：ENOENT（符合预期）
删除 workflow，测试已修改	通过
workflow 存在且有守卫，测试已修改	通过
workflow 存在但去掉 if: 守卫，测试已修改	失败：cursor-hygiene-webhook.yml must keep a job-level upstream identity boundary
mobile-android-apk.yml 去掉守卫，测试已修改	失败：mobile-android-apk.yml must keep a job-level upstream identity boundary
mobile-android-apk.yml 缺失，测试已修改	失败：ENOENT

所有临时改动均已还原。

本地测试结果（逐个运行）

命令	结果
node --test .github/scripts/opent3code-*.test.cjs	105/105 通过
node --test .github/scripts/*.test.cjs	152/152 通过
python3 .github/scripts/stage-preview-bundle.test.py	6/6 通过
node .github/scripts/opent3code-source-integrity.cjs --placeholder-only origin/main HEAD	通过
apps/server: vp test run src/provider/providerMaintenanceRunner	1 文件，18/18 通过
apps/server: vp test run src/provider/providerCompatibility.test.ts src/provider/ModelManifest.test.ts（含 #28 守卫）	2 文件，19/19 通过
apps/server: vp test run src/vcs/GitVcsDriverCore.test.ts src/storageCleanup.test.ts（自动合并文件）	2 文件，274/274 通过
vp lint .github/scripts/opent3code-release.test.cjs	通过（无诊断）
vp fmt --check（改动的测试文件与文档）	通过

未验证

• typecheck（未运行）
• 浏览器测试（未运行）
• PR Gate（等待 CI）
• 上游 web 端改动（AddEnvironmentDialog.tsx、ConnectionsSettings.tsx、Sidebar 相关）及 packages/client-runtime 的单元测试未在本地运行

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-10-21T18:44:49.948982Z	d8e659a	PR opened

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: d8e659a980

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Preserve the removal timeout for Windows preflight commands

On Windows, a non-forced linked-worktree removal now runs git ls-files and git status through this helper, but these calls omit timeoutMs and therefore inherit the driver's 30-second default. The previous git worktree remove path allowed five minutes, so a clean large worktree whose status scan takes 30–300 seconds will now fail before deletion. Pass the worktree-removal timeout through to these preflight executions as well.

Useful? React with 👍 / 👎.

[github-actions]

Thread transfer impact

✅ Thread transfer remains within every enforced ceiling.

Provider	Metric	Main baseline	This PR	Impact	PR ceiling
Codex	Total thread wire	5.0 KiB	5.0 KiB	−7 B (−0.1%)	6.8 KiB	✅
Codex	Thread snapshot wire	3.8 KiB	3.8 KiB	0 B (0.0%)	4.9 KiB	✅
Codex	Live turn WebSocket wire	1.2 KiB	1.2 KiB	−7 B (−0.6%)	2.0 KiB	✅
Codex	Live turn WebSocket decoded	20.9 KiB	20.9 KiB	0 B (0.0%)	29.3 KiB	✅
Codex	Live turn messages	2	2	0 (0.0%)	8	✅
Claude	Total thread wire	5.0 KiB	5.0 KiB	−7 B (−0.1%)	6.8 KiB	✅
Claude	Thread snapshot wire	3.8 KiB	3.8 KiB	0 B (0.0%)	4.9 KiB	✅
Claude	Live turn WebSocket wire	1.2 KiB	1.2 KiB	−7 B (−0.6%)	2.0 KiB	✅
Claude	Live turn WebSocket decoded	21.2 KiB	21.2 KiB	0 B (0.0%)	29.3 KiB	✅
Claude	Live turn messages	2	2	0 (0.0%)	8	✅

Baseline: b001e93 · PR result: d8e659a · Source CI: success

Scenario and decoded snapshot size

10 historical turns, 5 command tools per turn, 878.9 KiB retained MCP result per historical turn, and a 1.05 MiB retained result in the measured turn.

• Codex decoded thread snapshot: 108.5 KiB
• Claude decoded thread snapshot: 108.8 KiB

Updated in place by a trusted workflow. PR artifacts are strictly validated and never executed.