fix(templates): preserve local test policy

- Register testing discipline as merged during updates
- Keep project test commands in the local test skill
- Cover upgrade behavior and the watch self-heal contract

Co-Authored-By: Codex <noreply@openai.com>
Co-Authored-By: Claude <noreply@anthropic.com>

Author: 3 people

.agents/.airc.json

@@ -79,6 +79,7 @@
       "**/test-integration.*",
       "**/test.*",
       "**/upgrade-dependency.*",
+      ".agents/rules/testing-discipline.*",
       ".agents/QUICKSTART.md",
       ".agents/README.md",
       ".agents/skills/post-release/SKILL.*",

.agents/rules/testing-discipline.md

@@ -39,32 +39,40 @@ assert.match(content, /^name: code-task$/m);         // 正向断言已足够
 - **测试实现细节**：优先断言公开接口、产物、状态变化或错误结果；避免断言私有函数、内部调用顺序、临时数据结构。
 - **断言不充分**：断言必须锁定具体期望值；不要用"只要不抛异常""结果存在即可"替代对关键字段、数量和边界的验证。
 
-## 覆盖率定位（信息层）
+## 项目级测试策略
+
+仓库特定的命令、目录约定、覆盖率阈值、CI 集成和报告服务应记录在项目自己的测试文档中。新增测试前先读取该策略，并按测试的可观察范围与运行成本选择对应层级。
+
+如果项目没有定义分层测试套件，RED 与 GREEN 验证都使用项目的完整测试命令。不要在无关改动中自行引入测试层级或覆盖率门禁。
+
+### agent-infra 项目策略
+
+#### 覆盖率定位（信息层）
 
 > CI 中通过 `node --test --experimental-test-coverage` 输出覆盖率，仅作为"哪些文件被测试薄弱"的提示，**不作为 merge gate**。
 
-### 本地运行
+##### 本地运行
 
 ```bash
 npm run test:coverage
 ```
 
 stdout 末尾会打印按文件粒度的行 / 分支 / 函数覆盖率以及未覆盖行号。
 
-### CI 展示
+##### CI 展示
 
 `.github/workflows/unit-tests.yml` 在 ubuntu-latest 分片上把覆盖率块写入 GitHub Actions 的 step summary（PR Checks 页可见）。Windows / macOS 分片不重复输出。
 
 README 顶部的 Codecov 徽章由 `.github/workflows/unit-tests.yml` 在 ubuntu-latest 分片上传 `coverage.lcov` 后由 Codecov 生成。
 
-### 边界
+##### 边界
 
 - **不设置百分比阈值**：`--test-coverage-lines/branches/functions` 等阈值参数禁止加入；Goodhart's law 提醒我们一旦把覆盖率作为指标，开发者会写"覆盖率友好但行为弱"的测试。
 - **第三方服务仅用于徽章**：已接入 Codecov 托管 README 覆盖率徽章，但通过根 `codecov.yml` 显式关闭其 project/patch status check 与 PR 评论——Codecov 在本项目只展示数字，不参与 merge 决策。不接入 coveralls 等其他服务。
 - **不区分 tier**：当前只对 full `test` tier 输出覆盖率；smoke / core tier 的覆盖率没有独立价值。
 - **不阻塞 PR**：CI 步骤 `continue-on-error: true`，即便覆盖率采集失败也不影响 merge。
 
-### 新测试该放哪一层
+#### 新测试该放哪一层
 
 测试文件放入哪一层决定它会被哪些 npm script 自动执行：
 
@@ -74,7 +82,7 @@ README 顶部的 Codecov 徽章由 `.github/workflows/unit-tests.yml` 在 ubuntu
 
 模块继续作为第二级目录（如 `cli`、`core`、`scripts`、`templates`）。共享 helper 和 fixtures 保持在 `tests/helpers/`、`tests/helpers.ts`、`tests/fixtures/`，不要放入任一 tier。
 
-### 与"测试 tier 覆盖"的关系
+#### 与"测试 tier 覆盖"的关系
 
 注意区分两个概念：
 

.agents/skills/test/SKILL.md

@@ -11,11 +11,15 @@ description: >
 
 ## 1. 编译 / 类型检查
 
-本项目由 Node.js CLI 和模板文件组成，无需编译。跳过此步骤。
+```bash
+npm run typecheck
+```
+
+项目测试脚本会先运行 `npm run build`，因此单独执行类型检查后无需在本步骤重复构建。
 
 ## 2. 运行单元测试（按层级选择）
 
-本项目把测试分为三层，按场景选择运行命令；新增测试文件默认归入 **full**，确认足够快且足够核心后，再上调到 core 或 smoke。
+三层测试是反馈速度优化；本项目按测试的可观察范围与运行成本选择对应层级。新增测试文件默认归入 **full**，确认足够快且足够核心后，再上调到 core 或 smoke。
 
 ### smoke（目标 <5s）
 

.agents/skills/update-agent-infra/scripts/sync-templates.js

@@ -73,6 +73,7 @@ const DEFAULTS = {
       "**/test-integration.*",
       "**/test.*",
       "**/upgrade-dependency.*",
+      ".agents/rules/testing-discipline.*",
       ".agents/skills/post-release/SKILL.*",
       ".agents/skills/release/SKILL.*",
       ".agents/skills/test-integration/SKILL.*",

.agents/skills/watch-pr/reference/monitor-and-heal.md

@@ -12,6 +12,13 @@
 
 ## 自愈决策树
 
+```text
+# self-heal-test-command-contract
+primary: failing-job-command
+fallback-source: project-test-skill
+unknown: help
+```
+
 对每个失败 check，按下列顺序判定「自愈」还是「求助」：
 
 1. **能否定位到对应的 CI run**（按规则的「解析失败 run id」）？否 → 求助。
@@ -22,7 +29,7 @@
 4. 满足「可定位 + 代码层 + 未达上限」时执行一次自愈：
    - 自愈前先 `git status -s` 记录当前工作树，确保后续只纳入与本次失败相关的改动。
    - 在本地按日志定位并最小化修复（只动与该失败相关的代码 / 测试 / 配置）。
-   - 运行对应测试：优先失败 job 对应的本地命令；仓库统一回退 `npm run test:core`。**测试通过前不得提交或推送。**
+   - 运行对应测试：优先失败 job 对应的本地命令；否则读取项目 `test` skill，选择其中声明的 core 或 full 验证命令。两者均未知时进入求助出口。**测试通过前不得提交或推送。**
    - 测试通过后发布修复：按 `.agents/rules/commit-and-pr.md` 仅暂存本次相关文件（`git add <相关路径>`，避免 `git add -A` 卷入无关改动）→ 创建修复 commit（`git commit`）→ `git push` 到当前 PR 分支。
    - 记录本次修复的 commit SHA，修复计数 +1，回到 SKILL 步骤 2 重新监控。
    - 绝不执行与失败无关的「顺手优化」；不放宽 / 跳过失败的断言来「修绿」。

lib/defaults.json

@@ -52,6 +52,7 @@
       "**/test-integration.*",
       "**/test.*",
       "**/upgrade-dependency.*",
+      ".agents/rules/testing-discipline.*",
       ".agents/skills/post-release/SKILL.*",
       ".agents/skills/release/SKILL.*",
       ".agents/skills/test-integration/SKILL.*",

templates/.agents/skills/update-agent-infra/scripts/sync-templates.js

@@ -73,6 +73,7 @@ const DEFAULTS = {
       "**/test-integration.*",
       "**/test.*",
       "**/upgrade-dependency.*",
+      ".agents/rules/testing-discipline.*",
       ".agents/skills/post-release/SKILL.*",
       ".agents/skills/release/SKILL.*",
       ".agents/skills/test-integration/SKILL.*",

templates/.agents/skills/watch-pr/reference/monitor-and-heal.en.md

@@ -12,6 +12,13 @@ After running the watch command from `.agents/rules/pr-checks-commands.md`, clas
 
 ## Self-Heal Decision Tree
 
+```text
+# self-heal-test-command-contract
+primary: failing-job-command
+fallback-source: project-test-skill
+unknown: help
+```
+
 For each failing check, decide "self-heal" vs "ask for help" in this order:
 
 1. **Can the corresponding CI run be located** (per the rule's "Resolve a failing run id")? No → ask for help.
@@ -22,7 +29,7 @@ For each failing check, decide "self-heal" vs "ask for help" in this order:
 4. When "locatable + code layer + under cap" holds, perform one self-heal:
    - Before fixing, run `git status -s` to record the working tree and ensure only changes related to this failure are included.
    - Locate and make a minimal fix per the logs (touch only code / tests / config related to that failure).
-   - Run the relevant tests: prefer the local command for the failing job; otherwise use the project's declared full test command. If neither command is known, ask for help. **Do not commit or push before tests pass.**
+   - Run the relevant tests: prefer the local command for the failing job; otherwise read the project `test` skill and select its declared core or full validation command. If neither command is known, ask for help. **Do not commit or push before tests pass.**
    - After tests pass, publish the fix: per `.agents/rules/commit-and-pr.md` stage only the related files (`git add <paths>`, avoid `git add -A` pulling in unrelated changes) → create the fix commit (`git commit`) → `git push` to the current PR branch.
    - Record the fix commit SHA, increment the fix count, and return to SKILL step 2 to re-watch.
    - Never make unrelated "drive-by" optimizations; never loosen / skip the failing assertion to "make it green".

templates/.agents/skills/watch-pr/reference/monitor-and-heal.zh-CN.md

@@ -12,6 +12,13 @@
 
 ## 自愈决策树
 
+```text
+# self-heal-test-command-contract
+primary: failing-job-command
+fallback-source: project-test-skill
+unknown: help
+```
+
 对每个失败 check，按下列顺序判定「自愈」还是「求助」：
 
 1. **能否定位到对应的 CI run**（按规则的「解析失败 run id」）？否 → 求助。
@@ -22,7 +29,7 @@
 4. 满足「可定位 + 代码层 + 未达上限」时执行一次自愈：
    - 自愈前先 `git status -s` 记录当前工作树，确保后续只纳入与本次失败相关的改动。
    - 在本地按日志定位并最小化修复（只动与该失败相关的代码 / 测试 / 配置）。
-   - 运行对应测试：优先失败 job 对应的本地命令；否则使用项目已声明的完整测试命令。两者均未知时进入求助出口。**测试通过前不得提交或推送。**
+   - 运行对应测试：优先失败 job 对应的本地命令；否则读取项目 `test` skill，选择其中声明的 core 或 full 验证命令。两者均未知时进入求助出口。**测试通过前不得提交或推送。**
    - 测试通过后发布修复：按 `.agents/rules/commit-and-pr.md` 仅暂存本次相关文件（`git add <相关路径>`，避免 `git add -A` 卷入无关改动）→ 创建修复 commit（`git commit`）→ `git push` 到当前 PR 分支。
    - 记录本次修复的 commit SHA，修复计数 +1，回到 SKILL 步骤 2 重新监控。
    - 绝不执行与失败无关的「顺手优化」；不放宽 / 跳过失败的断言来「修绿」。

tests/unit/cli/sync-templates.test.ts

@@ -1099,3 +1099,60 @@ test("syncTemplates reports shared pre-commit as a merged pending file", async (
     fs.rmSync(tmpDir, { recursive: true, force: true });
   }
 });
+
+test("syncTemplates preserves project testing discipline as a merged file", async () => {
+  const originalExecSync = childProcess.execSync;
+  const tmpDir = fs.mkdtempSync(path.join(os.tmpdir(), "ai-collab-sync-testing-discipline-"));
+
+  try {
+    const projectRoot = path.join(tmpDir, "project");
+    const { templateRoot } = createTemplateInstall(tmpDir);
+    const target = ".agents/rules/testing-discipline.md";
+    const localPolicy = "Project-specific testing policy\n";
+
+    fs.mkdirSync(projectRoot, { recursive: true });
+    writeFile(templateRoot, ".agents/rules/testing-discipline.en.md", "Common testing discipline\n");
+    writeFile(templateRoot, ".agents/rules/testing-discipline.zh-CN.md", "通用测试纪律\n");
+    writeFile(projectRoot, target, localPolicy);
+    writeJson(projectRoot, ".agents/.airc.json", {
+      project: "demo",
+      org: "acme",
+      language: "zh-CN",
+      platform: { type: "github" },
+      files: {
+        managed: [".agents/rules/"],
+        merged: [],
+        ejected: []
+      }
+    });
+
+    childProcess.execSync = (command) => {
+      if (command === "git remote get-url origin") {
+        throw new Error("not a git repo");
+      }
+      throw new Error(`Unexpected command: ${command}`);
+    };
+
+    const { syncTemplates } = await loadFreshEsm<SyncTemplatesModule>(".agents/skills/update-agent-infra/scripts/sync-templates.js");
+    const firstReport = syncTemplates(projectRoot, templateRoot);
+    const secondReport = syncTemplates(projectRoot, templateRoot);
+
+    assert.ok(
+      firstReport.registryAdded.some(
+        (entry) => entry.entry === ".agents/rules/testing-discipline.*" && entry.list === "merged"
+      )
+    );
+    assert.equal(fs.readFileSync(path.join(projectRoot, target), "utf8"), localPolicy);
+    assert.ok(firstReport.managed.skippedMerged.includes(target));
+    assert.ok(secondReport.managed.skippedMerged.includes(target));
+    assert.deepEqual(
+      firstReport.merged.pending.filter((entry) => entry.target === target),
+      [{ target, template: ".agents/rules/testing-discipline.zh-CN.md" }]
+    );
+    assert.deepEqual(firstReport.managed.written, []);
+    assert.deepEqual(secondReport.managed.written, []);
+  } finally {
+    childProcess.execSync = originalExecSync;
+    fs.rmSync(tmpDir, { recursive: true, force: true });
+  }
+});