diff --git a/docs/reference/cli.md b/docs/reference/cli.md index c2bfb3da..215332c6 100644 --- a/docs/reference/cli.md +++ b/docs/reference/cli.md @@ -35,7 +35,7 @@ CLI 可以从 `master` 快速演进，但必须兼容 `deploy.json` 固定的 CI - `gh issue read [--repo owner/name] [--json body,title,state,comments]` 通过 GitHub REST 读取 issue title/body/state/url 和 comments，默认输出 JSON；`view` 只保留为兼容别名。兼容旧脚本的 `--json body` 和 `--json body,title,state,comments` 字段选择，且正文仍稳定暴露在 `.data.issue.body`，避免调用方因为 JSON 路径变化把空值当成正文。字段白名单是 `body,title,state,comments,number,url,author,createdAt,updatedAt`，未知字段必须结构化失败并带 `runnerDisposition=business-failed`。`gh issue create --title --body-file <file> [--label label[,label...]]... [--dry-run]`、`gh issue update <number> --mode replace|append --body-file <file> [--title ...] [--dry-run]`、`gh issue comment create <number> --body-file <file> [--dry-run]`、`gh issue comment delete <commentId> [--dry-run]`、`gh issue close|reopen <number> [--dry-run]` 都走 REST，不依赖 `gh` binary。`--label` 仅用于 `issue create`，支持重复传入和逗号分隔；`--dry-run` 会展示解析后的 labels 与 request plan，正式创建时把 labels 放入 GitHub REST create-issue payload，GitHub 返回不存在 label 等 422 校验失败时 CLI 结构化返回 `validation-failed`，不静默成功。`gh issue delete <number>` 是结构化 `unsupported-command`，因为 GitHub REST 不支持 issue 硬删除；生命周期删除语义请使用 `close`。 - `gh issue update <number> --mode replace|append --body-file <file>` 是正文更新主入口，`edit` 保留为兼容别名。`replace` 用文件正文替换现有 body；`append` 先读取当前 body，再按 UTF-8 文件字节追加，保留真实换行、反引号和 Markdown 表格。更新默认拒绝字面量 `null`、空白正文和过短正文；只有真实需要写短正文时才允许显式加 `--allow-short-body`，返回 JSON 会报告该风险。#20 总看板和指挥简报类 issue 是长期 body-only issue，`--body-profile auto` 会按 issue number 自动启用 #20/#24 legacy guard：#20 必须包含 `## 看板（OPEN）`，#24 legacy 指挥简报必须包含 `## 常驻观察与长期建议`。显式 `--body-profile commander-brief` 不再固定 #24；#24 仍兼容，标题为 `YYYY-MM-DD 指挥简报（北京时间）` 或既有正文首行/关键 heading 表明为每日滚动指挥简报的 issue 也合法，并仍必须包含 `## 常驻观察与长期建议`。对非简报 issue 显式使用 `commander-brief` 会结构化失败为 `profile-issue-mismatch`。`--dry-run` 不 PATCH GitHub，输出新正文长度、SHA、关键标题检查结果、字面量 `\n`、反引号、Markdown 表格和 shell 污染信号；若环境里有 `GH_TOKEN` 或 `GITHUB_TOKEN`，dry-run 还会只读抓取旧正文长度、SHA 和 `updatedAt` 作为更新前对照。正式写入可带 `--expect-updated-at <updated_at>` 或 `--expect-body-sha <sha256>`，CLI 会先读当前 issue，匹配后才 PATCH，防止旧缓存覆盖新正文。 - `gh issue edit 24 --body-file <file> --notify-claudeqq-brief-diff [--dry-run]` 是 legacy #24 指挥简报的通知入口。正式执行会先读取 GitHub 上 #24 旧正文并通过 #24 body profile guard，再从 `--body-file` 读取新正文；随后先 PATCH issue 主体，再把本次新增的 `## 更新 YYYY-MM-DD HH:MM 北京时间` 段落发送给 ClaudeQQ，ClaudeQQ 失败不会回滚 issue 正文，失败只体现在返回 JSON 的 `claudeqq.ok=false` 和结构化 `degradedReason`。每日滚动简报 issue 可用普通 `gh issue update <number> --body-profile commander-brief --dry-run` 和并发 guard 更新，但此通知 helper 仍只支持 #24。带通知 flag 的 `--dry-run` 不 PATCH、不发送；它按新正文做发送预览，并在输出中标明非 dry-run 才会读取旧正文做可靠 diff。默认 ClaudeQQ 目标是私聊 `645275593`，默认 base URL 是 UniDesk 受控入口 `http://backend-core:8080/api/microservices/claudeqq/proxy`，可用 `UNIDESK_COMMANDER_BRIEF_CLAUDEQQ_ENABLED`、`UNIDESK_COMMANDER_BRIEF_CLAUDEQQ_BASE_URL`、`UNIDESK_COMMANDER_BRIEF_CLAUDEQQ_TARGET_TYPE`、`UNIDESK_COMMANDER_BRIEF_CLAUDEQQ_USER_ID`、`UNIDESK_COMMANDER_BRIEF_CLAUDEQQ_GROUP_ID` 和 `UNIDESK_COMMANDER_BRIEF_CLAUDEQQ_TIMEOUT_MS` 覆盖。 -- `gh issue board-audit [--repo owner/name] [--board-issue 20] [--limit N] [--known-meta-issue N[,N...]] [--ignore-issue N[,N...]] [--dry-run]` 是 #20 长期总看板只读覆盖审计入口，默认 repo 为 `pikasTech/unidesk`、board issue 为 `20`、输出 JSON 且不 PATCH/POST/DELETE GitHub。它读取 board body、GitHub open issue 列表和 closed issue 列表，对比 OPEN/CLOSED Markdown 表格并输出 `missingOpenIssues`、`closedInOpenRows`、`missingClosedRows`、`openInClosedRows`、`rowValidationWarnings`、`ignoredIssues` 和 `recommendedActions`。当表格里存在 Issue 列时，row.issueNumber 优先取 Issue cell 中第一个指向 `/issues/<N>` 的 Markdown link，找不到时取开头的 `#N`；同一 Issue cell 里主引用后面的标题说明引用（例如 `#20 总看板`、`基于 #4`）不触发 `multiple-issue-references`。没有 Issue 列的旧表格仍回退到整行 issue 提取，并保留多 issue 引用告警。`相关 Code Queue 任务`/`relatedTask` 列允许 `—`、`-`、`n/a`、`无任务` 等无关联任务占位表示 closed 历史/治理项没有 Code Queue task；这个放宽不适用于 branch、acceptance 或 progress。默认把 #20 和 #24 作为 `known-meta` 治理/简报 issue 忽略；需要扩展治理项用 `--known-meta-issue`，临时排除业务 issue 用 `--ignore-issue`。指挥官发现总看板可能漏行时，应先跑 board audit 获取结构化结果，再决定是否人工编辑 #20，而不是只靠 grep。 +- `gh issue board-audit [--repo owner/name] [--board-issue 20] [--limit N] [--known-meta-issue N[,N...]] [--ignore-issue N[,N...]] [--dry-run]` 是 #20 长期总看板只读覆盖审计入口，默认 repo 为 `pikasTech/unidesk`、board issue 为 `20`、输出 JSON 且不 PATCH/POST/DELETE GitHub。它读取 board body、GitHub open issue 列表和 closed issue 列表，对比 OPEN/CLOSED Markdown 表格并输出 `missingOpenIssues`、`closedInOpenRows`、`missingClosedRows`、`openInClosedRows`、`rowValidationWarnings`、`ignoredIssues` 和 `recommendedActions`。当表格里存在 Issue 列时，row.issueNumber 优先取 Issue cell 中第一个指向 `/issues/<N>` 的 Markdown link，找不到时取开头的 `#N`；同一 Issue cell 里主引用后面的标题说明引用（例如 `#20 总看板`、`基于 #4`）不触发 `multiple-issue-references`。没有 Issue 列的旧表格仍回退到整行 issue 提取，并保留多 issue 引用告警。`相关 Code Queue 任务`/`relatedTask` 列允许 `—`、`-`、`n/a`、`无任务` 等无关联任务占位表示 closed 历史/治理项没有 Code Queue task；这个放宽不适用于 branch、acceptance 或 progress。默认把 #20 和 #24 作为 `known-meta` 治理/简报 issue 忽略；标题匹配 `YYYY-MM-DD 指挥简报（北京时间）` 的每日滚动简报由 #20 顶部指挥简报索引管理，不进入 OPEN/CLOSED 覆盖审计，并在 `ignoredIssues` 中标记 `reason=brief-index-managed`。需要扩展治理项用 `--known-meta-issue`，临时排除业务 issue 用 `--ignore-issue`。指挥官发现总看板可能漏行时，应先跑 board audit 获取结构化结果，再决定是否人工编辑 #20，而不是只靠 grep。 - `gh issue board-row list --board-issue 20 [--state open|closed|all] [--dry-run]`、`gh issue board-row get <issueNumber> --board-issue 20` 和 `gh issue board-row update <issueNumber> --board-issue 20 --field progress|status|validation|branch|tasks|focus --value <text> [--dry-run] [--expect-updated-at ts|--expect-body-sha sha256]` 是 #20 看板表格单行结构化入口。list/get 复用 board-audit parser，只读返回 row、cells、fields、section、lineNumber、bodySha 和 rowValidationWarnings。update 只替换命中的一行里一个单元格，返回 old/new row、old/new body SHA、body guard、request plan 和 parser 结果；默认没有并发期望时即使不写 `--dry-run` 也只做 dry-run，正式 PATCH 必须带 `--expect-body-sha` 或 `--expect-updated-at`。字段映射固定为：`branch` -> Branch，`progress` -> 进度，`status`/`validation` -> 验收状态，`tasks` -> 相关 Code Queue 任务，`focus` -> 当前关注点。单元格值中的 Markdown 表格管道会转义为 `\|`，真实换行会折叠为空格，避免新增字面量 `\n` 污染。`gh issue board-row add <issueNumber> --board-issue 20 --section open|closed --row-file <file> [--dry-run] [--expect-body-sha|--expect-updated-at]`、`move <issueNumber> --board-issue 20 --to open|closed [--status OPEN|CLOSED] [--dry-run] [--expect-body-sha|--expect-updated-at]` 和 `delete <issueNumber> --board-issue 20 [--dry-run] [--expect-body-sha|--expect-updated-at]` 是 row-scoped #20 结构化写入口。add 校验一行 `--row-file` 的 Issue 列、列数和 GitHub 状态列与目标 section 一致；move 允许跨 OPEN/CLOSED 表迁移并在需要时同步 GitHub 状态列；delete 仅删除匹配行。三者默认 dry-run，非 dry-run 必须带 `--expect-body-sha` 或 `--expect-updated-at`，并返回 old/new row、body SHA、line/section 计划和 parser 结果。 - `gh issue scan-escape [--repo owner/name] [--limit N] [--dry-run]` 只读扫描 issue 主体和 comments 中的字面量 `\n`、可疑 `\t`、shell newline escape、escaped backtick、ANSI escape 字符串、短 body、blank body 和 null body。输出固定 JSON，`findings` 会带 `bodyKind=issue-body|comment-body`、`issueNumber`、`issueId`、`commentId`、`lineNumber`、`column`、`kind`、`snippet` 和 `classification=suspected-pollution|explanatory-mention|risk`，用于区分说明性提到 `\n` 和疑似污染；`cleanupSuggestions` 只给 dry-run 清理建议、body/comment 定位和 diff-like preview，不 PATCH、不 DELETE、不真实清理历史 comment。`gh issue cleanup-plan` 是同一只读能力的别名，默认 `dryRun=true`。`gh pr list|read|view [--json ...]` 提供 REST 列表和详情，PR 字段白名单是 `body,title,state,number,url,author,head,base,draft,createdAt,updatedAt`。`gh pr create --title <title> --body-file <file>|--body <text> --base <branch> --head <branch> [--draft] [--dry-run]`、`gh pr update <number> --mode replace|append --body-file <file>|--body <text> [--title ...] [--dry-run]`、`gh pr comment create <number> --body-file <file>|--body <text> [--dry-run]`、`gh pr comment delete <commentId> [--dry-run]`、`gh pr close|reopen <number> [--dry-run]` 是 PR CRUD/生命周期入口。`pr create --dry-run` 只输出 planned operation，不访问 GitHub；非 dry-run 创建前会校验 repo、base、head 和 compare ahead 状态，成功时返回 PR number/url。`pr update --mode append` 会先读取当前 PR body 再追加正文。`gh pr delete <number>` 和 `gh pr merge` 本阶段不开放，始终结构化返回 `unsupported-command`；PR 生命周期删除语义请使用 `close`。 - PR dry-run/probe 的最小手动序列是：`bun scripts/cli.ts gh auth status --repo pikasTech/unidesk` 只读检查 token 来源、GitHub REST egress、repo 可见性和 issue read；`bun scripts/cli.ts gh pr create --repo pikasTech/unidesk --title <title> --body-file <file> --base master --head <head> --dry-run` 检查创建计划；`bun scripts/cli.ts gh pr list --repo pikasTech/unidesk --limit 5 --json number,title,state,url,head,base` 和 `bun scripts/cli.ts gh pr read <number> --repo pikasTech/unidesk --json body,title,state,head,base` 做只读 PR 观察；`bun scripts/cli.ts gh pr comment <number> --repo pikasTech/unidesk --body-file <file> --dry-run` 检查评论计划；`bun scripts/cli.ts gh pr merge <number> --repo pikasTech/unidesk` 必须失败并返回结构化 `unsupported-command`。Code Queue runner 可用 `bun scripts/code-queue-pr-preflight-example.ts --repo pikasTech/unidesk --base master --head <head> --comment-pr <number>` 一次性跑只读 auth status 与 PR create/comment dry-run；该脚本不得输出 token 值，也不会创建、评论或 merge PR。 diff --git a/docs/reference/code-queue-supervision.md b/docs/reference/code-queue-supervision.md index 82ba1eb4..3f5ef995 100644 --- a/docs/reference/code-queue-supervision.md +++ b/docs/reference/code-queue-supervision.md @@ -95,7 +95,7 @@ issue 内容必须自包含，至少写清楚背景、外部收益、当前观 如果某个 worker 任务需要依赖 GitHub issue 内容，但 runner 的 issue 可达性尚未被单独验证，指挥官不能默认 worker 已能读取该 issue。此时 worker prompt 必须直接内嵌完整需求、约束和验收点，issue URL 只能作为辅助引用。若要把 issue 作为任务输入源，先单独做可达性探测，再决定是否把 issue 作为常规前置条件。 -GitHub issue/PR 操作应优先使用 UniDesk CLI 的安全入口：`bun scripts/cli.ts gh auth status`、`gh issue list/read/view/create/update/comment create/comment delete/close/reopen/scan-escape/cleanup-plan/board-audit/board-row list/board-row get/board-row update`、`gh pr list/read/view/create/update/comment create/comment delete/close/reopen`。该入口默认 repo 是 `pikasTech/unidesk`，支持 `--repo owner/name`，输出稳定 JSON，并把 `missing-binary`、`missing-token`、`auth-failed`、`network-proxy-failed`、`permission-denied`、`repo-not-found`、`repo-forbidden`、`issue-not-found`、`pr-not-found`、`scope-insufficient`、`validation-failed`、`invalid-response`、`unsupported-command` 等失败原因结构化。失败对象必须包含 `runnerDisposition=infra-blocked|business-failed`，runner 应用它区分基础设施阻塞和业务/参数失败。`gh issue list --state open --limit N --json number,title,state,url` 是有界 issue 发现入口，`--state` 只接受 `open|closed|all`，list 字段白名单是 `number,title,state,url,updatedAt,createdAt,author,labels`；未知 state 或未知字段必须失败，不能静默返回空数组。`gh issue read <number> --json body` 是 canonical 入口，正文仍应从 `.data.issue.body` 读取；`view` 只保留为兼容别名。未知 `--json` 字段必须失败，不得让调用方把空正文误判为读取成功。`gh issue scan-escape --limit N [--dry-run]` 与 `gh issue cleanup-plan` 只读扫描 issue body/comments 的字面量 `\n`、shell escape、短 body、blank/null body，输出 `classification=suspected-pollution|explanatory-mention|risk`、body/comment id、预览和清理建议；说明性提到 `\n` 不应被当成污染，cleanup-plan 永远不真实清理历史评论。`gh issue board-audit --board-issue 20 --limit N --dry-run` 只读对比 GitHub open/closed issue 列表与 #20 的 OPEN/CLOSED Markdown 表格，输出 `missingOpenIssues`、`closedInOpenRows`、`missingClosedRows`、`rowValidationWarnings`、`ignoredIssues` 和 `recommendedActions`。存在 Issue 列时，row key 优先来自 Issue cell 的主 Markdown issue link `/issues/<N>`，其次是开头 `#N`；主引用后的标题说明引用不触发多 issue 告警。没有 Issue 列的旧表格仍使用整行 fallback 并保留 `multiple-issue-references`。`相关 Code Queue 任务` 可用 `—`、`-` 或等价无任务占位表达无关联 task，但 branch、acceptance 和 progress 仍必须填写真实值；默认把 #20/#24 作为 known-meta 忽略。指挥官发现 #20 可能漏跟 open issue 或分表错误时，应先跑 board audit 取得结构化缺口，再人工审阅和编辑 #20，不要只靠 grep。显式 `gh issue update <number> --body-profile commander-brief` 可用于 #24 legacy 简报和每日滚动简报 issue；每日简报 issue 应用标题 `YYYY-MM-DD 指挥简报（北京时间）` 或在既有正文首行/关键 heading 中标明简报身份，且新正文必须包含 `## 常驻观察与长期建议`。对非简报 issue 使用该 profile 应失败为 `profile-issue-mismatch`。`gh issue board-row list --board-issue 20 --state open|closed|all` 和 `gh issue board-row get <issueNumber> --board-issue 20` 复用同一个 parser 读取表格行；`gh issue board-row update <issueNumber> --board-issue 20 --field progress|status|validation|branch|tasks|focus --value <text>` 只替换一行一个单元格，输出 old/new row、body SHA、body guard 和 request plan，且默认 dry-run，正式写入必须带 `--expect-body-sha` 或 `--expect-updated-at`。字段映射中 `status`/`validation` 都指向 `验收状态`，`tasks` 指向 `相关 Code Queue 任务`，`focus` 指向 `当前关注点`；单元格管道会转义、真实换行会折叠为空格，避免新增字面量 `\n`。`board-row move/delete` 暂时结构化 unsupported。`gh pr list|read|view --json ...` 支持 `body,title,state,number,url,author,head,base,draft,createdAt,updatedAt` 字段白名单。issue/PR 创建、更新、评论、评论删除、关闭和重开使用 GitHub REST API；只要有 `GH_TOKEN` 或 `GITHUB_TOKEN`，就不依赖系统 `gh` binary。`gh` binary 只作为状态探测和 `gh auth token` fallback，不是写操作的主路径。GitHub 不支持 issue/PR 硬删除，`gh issue delete` 和 `gh pr delete` 必须结构化返回 `unsupported-command`；生命周期删除语义使用 `close`。`gh pr merge` 仍然不开放。 +GitHub issue/PR 操作应优先使用 UniDesk CLI 的安全入口：`bun scripts/cli.ts gh auth status`、`gh issue list/read/view/create/update/comment create/comment delete/close/reopen/scan-escape/cleanup-plan/board-audit/board-row list/board-row get/board-row update`、`gh pr list/read/view/create/update/comment create/comment delete/close/reopen`。该入口默认 repo 是 `pikasTech/unidesk`，支持 `--repo owner/name`，输出稳定 JSON，并把 `missing-binary`、`missing-token`、`auth-failed`、`network-proxy-failed`、`permission-denied`、`repo-not-found`、`repo-forbidden`、`issue-not-found`、`pr-not-found`、`scope-insufficient`、`validation-failed`、`invalid-response`、`unsupported-command` 等失败原因结构化。失败对象必须包含 `runnerDisposition=infra-blocked|business-failed`，runner 应用它区分基础设施阻塞和业务/参数失败。`gh issue list --state open --limit N --json number,title,state,url` 是有界 issue 发现入口，`--state` 只接受 `open|closed|all`，list 字段白名单是 `number,title,state,url,updatedAt,createdAt,author,labels`；未知 state 或未知字段必须失败，不能静默返回空数组。`gh issue read <number> --json body` 是 canonical 入口，正文仍应从 `.data.issue.body` 读取；`view` 只保留为兼容别名。未知 `--json` 字段必须失败，不得让调用方把空正文误判为读取成功。`gh issue scan-escape --limit N [--dry-run]` 与 `gh issue cleanup-plan` 只读扫描 issue body/comments 的字面量 `\n`、shell escape、短 body、blank/null body，输出 `classification=suspected-pollution|explanatory-mention|risk`、body/comment id、预览和清理建议；说明性提到 `\n` 不应被当成污染，cleanup-plan 永远不真实清理历史评论。`gh issue board-audit --board-issue 20 --limit N --dry-run` 只读对比 GitHub open/closed issue 列表与 #20 的 OPEN/CLOSED Markdown 表格，输出 `missingOpenIssues`、`closedInOpenRows`、`missingClosedRows`、`rowValidationWarnings`、`ignoredIssues` 和 `recommendedActions`。存在 Issue 列时，row key 优先来自 Issue cell 的主 Markdown issue link `/issues/<N>`，其次是开头 `#N`；主引用后的标题说明引用不触发多 issue 告警。没有 Issue 列的旧表格仍使用整行 fallback 并保留 `multiple-issue-references`。`相关 Code Queue 任务` 可用 `—`、`-` 或等价无任务占位表达无关联 task，但 branch、acceptance 和 progress 仍必须填写真实值；默认把 #20/#24 作为 known-meta 忽略，标题匹配 `YYYY-MM-DD 指挥简报（北京时间）` 的每日滚动简报由 #20 顶部指挥简报索引管理，并以 `reason=brief-index-managed` 进入 `ignoredIssues`，不进入 OPEN/CLOSED 覆盖审计。指挥官发现 #20 可能漏跟 open issue 或分表错误时，应先跑 board audit 取得结构化缺口，再人工审阅和编辑 #20，不要只靠 grep。显式 `gh issue update <number> --body-profile commander-brief` 可用于 #24 legacy 简报和每日滚动简报 issue；每日简报 issue 应用标题 `YYYY-MM-DD 指挥简报（北京时间）` 或在既有正文首行/关键 heading 中标明简报身份，且新正文必须包含 `## 常驻观察与长期建议`。对非简报 issue 使用该 profile 应失败为 `profile-issue-mismatch`。`gh issue board-row list --board-issue 20 --state open|closed|all` 和 `gh issue board-row get <issueNumber> --board-issue 20` 复用同一个 parser 读取表格行；`gh issue board-row update <issueNumber> --board-issue 20 --field progress|status|validation|branch|tasks|focus --value <text>` 只替换一行一个单元格，输出 old/new row、body SHA、body guard 和 request plan，且默认 dry-run，正式写入必须带 `--expect-body-sha` 或 `--expect-updated-at`。字段映射中 `status`/`validation` 都指向 `验收状态`，`tasks` 指向 `相关 Code Queue 任务`，`focus` 指向 `当前关注点`；单元格管道会转义、真实换行会折叠为空格，避免新增字面量 `\n`。`board-row move/delete` 暂时结构化 unsupported。`gh pr list|read|view --json ...` 支持 `body,title,state,number,url,author,head,base,draft,createdAt,updatedAt` 字段白名单。issue/PR 创建、更新、评论、评论删除、关闭和重开使用 GitHub REST API；只要有 `GH_TOKEN` 或 `GITHUB_TOKEN`，就不依赖系统 `gh` binary。`gh` binary 只作为状态探测和 `gh auth token` fallback，不是写操作的主路径。GitHub 不支持 issue/PR 硬删除，`gh issue delete` 和 `gh pr delete` 必须结构化返回 `unsupported-command`；生命周期删除语义使用 `close`。`gh pr merge` 仍然不开放。 CLI 是短 shout 的需求原语，不是长驻服务器进程。CLI 功能不好用、兼容性不足、安全 guard 不够或输出不利于 runner/指挥官使用时，应默认创建 GitHub issue 并用 Code Queue 推进；这类 CLI 问题走 `master`、remote commit、轻量 contract test 和文档更新，不套用 backend-core、Code Queue runtime 这类运行态服务的重部署门禁。若 CLI 缺陷已经阻塞当前指挥，可以先做最小安全绕行，同时把长期修复写入 issue 并派单。 diff --git a/scripts/gh-cli-issue-guard-contract-test.ts b/scripts/gh-cli-issue-guard-contract-test.ts index 6942e6c9..9269a9f0 100644 --- a/scripts/gh-cli-issue-guard-contract-test.ts +++ b/scripts/gh-cli-issue-guard-contract-test.ts @@ -297,6 +297,7 @@ async function startMockGitHub(): Promise<{ baseUrl: string; requests: MockReque created_at: "2026-05-20T01:30:00Z", updated_at: "2026-05-20T02:30:00Z", }, + dailyCommanderBriefIssue, ]; const legacyBoardOpenIssues = [ { @@ -603,6 +604,7 @@ export async function runGhCliIssueGuardContract(): Promise<JsonRecord> { const missingOpenIssues = boardAuditData.missingOpenIssues as JsonRecord[]; assertCondition(Array.isArray(missingOpenIssues) && missingOpenIssues.some((issue) => issue.number === 36), "board-audit should report open issue missing from OPEN table", missingOpenIssues); assertCondition(!missingOpenIssues.some((issue) => issue.number === 45), "board-audit should use the primary markdown issue link as the Issue-column row key", missingOpenIssues); + assertCondition(!missingOpenIssues.some((issue) => issue.number === 46), "board-audit should not require daily commander briefs in the OPEN table", missingOpenIssues); const closedInOpenRows = boardAuditData.closedInOpenRows as JsonRecord[]; assertCondition(Array.isArray(closedInOpenRows) && closedInOpenRows.some((issue) => issue.number === 40), "board-audit should report closed issue still in OPEN table", closedInOpenRows); const missingClosedRows = boardAuditData.missingClosedRows as JsonRecord[]; @@ -618,8 +620,10 @@ export async function runGhCliIssueGuardContract(): Promise<JsonRecord> { assertCondition(!rowValidationWarnings.some((warning) => Array.isArray(warning.missingColumns) && (warning.missingColumns as unknown[]).includes("relatedTask")), "board-audit should recognize relatedTask from the current header", rowValidationWarnings); const ignoredIssues = boardAuditData.ignoredIssues as JsonRecord[]; assertCondition(Array.isArray(ignoredIssues) && ignoredIssues.some((issue) => issue.number === 20 && issue.reason === "known-meta") && ignoredIssues.some((issue) => issue.number === 24 && issue.reason === "known-meta"), "board-audit should ignore configured known meta issues", ignoredIssues); + assertCondition(ignoredIssues.some((issue) => issue.number === 46 && issue.reason === "brief-index-managed"), "board-audit should ignore daily commander briefs managed by the #20 brief index", ignoredIssues); const recommendedActions = boardAuditData.recommendedActions as JsonRecord[]; assertCondition(Array.isArray(recommendedActions) && recommendedActions.some((action) => action.action === "add-open-row" && action.issueNumber === 36) && recommendedActions.some((action) => action.action === "move-open-row-to-closed" && action.issueNumber === 40), "board-audit should emit machine-readable recommended actions", recommendedActions); + assertCondition(!recommendedActions.some((action) => action.action === "add-open-row" && action.issueNumber === 46), "board-audit should not recommend adding daily commander briefs to the OPEN table", recommendedActions); const boardAuditWriteCount = mock.requests.slice(boardAuditRequestCountBefore).filter((request) => request.method === "PATCH" || request.method === "DELETE" || request.method === "POST").length; assertCondition(boardAuditWriteCount === 0, "board-audit must not write GitHub", { requests: mock.requests.slice(boardAuditRequestCountBefore) }); @@ -1072,7 +1076,7 @@ export async function runGhCliIssueGuardContract(): Promise<JsonRecord> { "issue list default fields include labels and filter pull requests", "issue scan-escape classifies pollution, explanatory mentions, and body risks", "issue cleanup-plan remains dry-run with body/comment cleanup suggestions", - "issue board-audit reports missing open rows, closed/open section mismatches, missing closed rows, meta ignores, and row validation warnings without writes", + "issue board-audit reports missing open rows, closed/open section mismatches, missing closed rows, meta/brief-index ignores, and row validation warnings without writes", "issue board-row list/get expose parsed #20 rows without writes", "issue board-row add/delete without guard stay on dry-run and do not PATCH", "issue board-row update defaults to dry-run, reports old/new row, body SHA, guard result, and does not introduce literal backslash-n", diff --git a/scripts/src/gh.ts b/scripts/src/gh.ts index 495db18d..6b52b96b 100644 --- a/scripts/src/gh.ts +++ b/scripts/src/gh.ts @@ -58,6 +58,7 @@ type BoardSectionKind = "open" | "closed"; type BoardRequiredColumn = typeof BOARD_AUDIT_REQUIRED_COLUMNS[number]; type BoardColumnKind = BoardRequiredColumn | "focus" | "githubStatus"; type BoardRowField = typeof BOARD_ROW_FIELDS[number]; +type BoardIgnoreReason = "known-meta" | "ignored" | "brief-index-managed"; type GitHubDegradedReason = | "missing-binary" @@ -218,7 +219,7 @@ interface BoardIssueEntry { interface BoardIgnoredIssue { number: number; - reason: "known-meta" | "ignored"; + reason: BoardIgnoreReason; title?: string; state?: string; url?: string; @@ -1285,10 +1286,15 @@ function mergedKnownMetaIssues(options: GitHubOptions): number[] { return values; } -function boardIgnoreMap(options: GitHubOptions): Map<number, "known-meta" | "ignored"> { - const ignored = new Map<number, "known-meta" | "ignored">(); +function boardIgnoreMap(options: GitHubOptions, issues: BoardIssueEntry[] = []): Map<number, BoardIgnoreReason> { + const ignored = new Map<number, BoardIgnoreReason>(); for (const issueNumber of mergedKnownMetaIssues(options)) ignored.set(issueNumber, "known-meta"); for (const issueNumber of options.ignoredIssues) ignored.set(issueNumber, "ignored"); + if (options.boardIssue === CODE_QUEUE_BOARD_TARGET_ISSUE) { + for (const issue of issues) { + if (!ignored.has(issue.number) && isDailyCommanderBriefTitle(issue.title)) ignored.set(issue.number, "brief-index-managed"); + } + } return ignored; } @@ -2540,7 +2546,7 @@ async function issueBoardRowDelete(repo: string, token: string, issueNumber: num ); } -function ignoredIssueList(issues: BoardIssueEntry[], ignoreMap: Map<number, "known-meta" | "ignored">): BoardIgnoredIssue[] { +function ignoredIssueList(issues: BoardIssueEntry[], ignoreMap: Map<number, BoardIgnoreReason>): BoardIgnoredIssue[] { return sortedIssueEntries(issues.filter((issue) => ignoreMap.has(issue.number))).map((issue) => ({ number: issue.number, reason: ignoreMap.get(issue.number) ?? "ignored", @@ -2550,7 +2556,7 @@ function ignoredIssueList(issues: BoardIssueEntry[], ignoreMap: Map<number, "kno })); } -function ignoredBoardOnlyIssues(issueNumbers: number[], issueMap: Map<number, BoardIssueEntry>, ignoreMap: Map<number, "known-meta" | "ignored">): BoardIgnoredIssue[] { +function ignoredBoardOnlyIssues(issueNumbers: number[], issueMap: Map<number, BoardIssueEntry>, ignoreMap: Map<number, BoardIgnoreReason>): BoardIgnoredIssue[] { return issueNumbers .filter((issueNumber) => ignoreMap.has(issueNumber) && !issueMap.has(issueNumber)) .sort((a, b) => a - b) @@ -3174,7 +3180,7 @@ async function issueBoardAudit(repo: string, token: string, options: GitHubOptio const closedIssues = rawClosedIssues.filter((issue) => issue.pull_request === undefined).slice(0, options.limit).map(boardIssueEntry); const allListedIssues = [...openIssues, ...closedIssues]; const issueMap = new Map<number, BoardIssueEntry>(allListedIssues.map((issue) => [issue.number, issue])); - const ignoreMap = boardIgnoreMap(options); + const ignoreMap = boardIgnoreMap(options, allListedIssues); const parsed = parseBoardTables(boardIssue.body ?? ""); const rowValidationWarnings = parsed.warnings.filter((warning) => warning.issueNumber === null || !ignoreMap.has(warning.issueNumber)); const openRows = parsed.sections.filter((section) => section.kind === "open").flatMap((section) => section.rows); @@ -3278,6 +3284,7 @@ async function issueBoardAudit(repo: string, token: string, options: GitHubOptio limit: options.limit, knownMetaIssues: mergedKnownMetaIssues(options), ignoredIssues: options.ignoredIssues, + briefIndexManagedIssuePattern: options.boardIssue === CODE_QUEUE_BOARD_TARGET_ISSUE ? "YYYY-MM-DD 指挥简报（北京时间）" : null, requiredColumns: BOARD_AUDIT_REQUIRED_COLUMNS.slice(), }, summary: { @@ -4092,7 +4099,7 @@ export function ghHelp(): unknown { "When staging a body file from a shell, use a quoted heredoc such as cat <<'EOF' > /tmp/body.md so backticks and backslashes are not expanded before --body-file reads the file.", "For JSON request bodies in other CLI namespaces, prefer --body-file or --body-stdin over long inline shell arguments; GitHub Markdown writes intentionally use --body-file only.", "issue scan-escape classifies literal \\n findings as suspected-pollution, explanatory-mention, or risk, and emits cleanupSuggestions with body/comment ids plus diff-like previews. cleanup-plan is an alias that remains dry-run/read-only.", - "issue board-audit is read-only and defaults to repo pikasTech/unidesk plus board issue #20. It compares GitHub open/closed issue lists with the board OPEN/CLOSED tables and reports missingOpenIssues, closedInOpenRows, missingClosedRows, rowValidationWarnings, ignoredIssues, and recommendedActions. When an Issue column exists, row.issueNumber is taken from that column; #20 and #24 are known meta issues by default.", + "issue board-audit is read-only and defaults to repo pikasTech/unidesk plus board issue #20. It compares GitHub open/closed issue lists with the board OPEN/CLOSED tables and reports missingOpenIssues, closedInOpenRows, missingClosedRows, rowValidationWarnings, ignoredIssues, and recommendedActions. When an Issue column exists, row.issueNumber is taken from that column; #20 and #24 are known meta issues by default. Daily rolling brief issues titled YYYY-MM-DD 指挥简报（北京时间） are managed by #20's brief index and appear in ignoredIssues with reason=brief-index-managed instead of missingOpenIssues.", "issue board-row list/get reuse the board-audit table parser and are read-only. board-row update changes one table cell by issue number, returns old/new row, body SHA, body guard and request plan, and defaults to dry-run unless --expect-updated-at or --expect-body-sha is supplied for the guarded PATCH. Field aliases map status and validation to the 验收状态 column, tasks to 相关 Code Queue 任务, and focus to 当前关注点.", "issue board-row add/move/delete are row-scoped #20 table mutations. add validates a one-line --row-file against the target table column count, Issue column, and GitHub 状态 column; move refuses duplicate/ambiguous rows and can update GitHub 状态 via --status; delete removes only the matched row. All three default to dry-run and require --expect-body-sha or --expect-updated-at before PATCH. add/move/delete return old/new row, body SHA, and line/section plan details for the parsed table mutation.", "issue edit 24 --notify-claudeqq-brief-diff remains the legacy #24 notification helper: it reads the old issue body, PATCHes the new body, and sends only newly added '## 更新 ... 北京时间' sections to ClaudeQQ; ClaudeQQ failure does not roll back GitHub.",