daily-standup
TeamBrain repo **24h window delta** report — 拉过去 N 小时窗口内 PR/issue 的创建/合并/关闭事件,按主题(features/bugfixes/docs/infra)分组,鸭语汇报。仅当用户问 `what happened since yesterday` / `explore today's issues and PRs` / `daily standup` / `/daily-standup` / `今天的 issue 和 PR` / `昨天到今天发生了啥` / `show me yesterday's activity` / `过去 24 小时 GitHub 活动` / `9 点到 9 点发生了什么` 这类**时间窗口 delta 类**问题时触发。**不要在以下情况触发**:(a) 用户问当前快照(`status of all issues` / `所有 open issue` → 用 `repo-issues-status` skill);(b) 用户问今日开工晨报(`/newday` → 用 `newday` skill,覆盖 worktree + 当前开放工作而非 24h delta);(c) 用户问单 issue/PR 状态(`#330 怎么样` → 用 `gh issue view`/`gh pr view`)。匹配命中后默认拉 yesterday 09:00 +0800 → today 09:00 +0800 的窗口,事件按 PRs merged / created / closed-without-merge / 还开着 + Issues created / closed / 还开着 六桶分组,再按主题二次归类,全文用 呷呷~ 鸭鸭 口吻包裹。
- License 许可
- MIT
- 最近更新
- 2026-05-20
- Stars
- ★ 1
- Forks
- 0
01 前言 Front Matter
- name
- daily-standup
- description
- TeamBrain repo **24h window delta** report — 拉过去 N 小时窗口内 PR/issue 的创建/合并/关闭事件,按主题(features/bugfixes/docs/infra)分组,鸭语汇报。仅当用户问 `what happened since yesterday` / `explore today's issues and PRs` / `daily standup` / `/daily-standup` / `今天的 issue 和 PR` / `昨天到今天发生了啥` / `show me yesterday's activity` / `过去 24 小时 GitHub 活动` / `9 点到 9 点发生了什么` 这类**时间窗口 delta 类**问题时触发。**不要在以下情况触发**:(a) 用户问当前快照(`status of all issues` / `所有 open issue` → 用 `repo-issues-status` skill);(b) 用户问今日开工晨报(`/newday` → 用 `newday` skill,覆盖 worktree + 当前开放工作而非 24h delta);(c) 用户问单 issue/PR 状态(`#330 怎么样` → 用 `gh issue view`/`gh pr view`)。匹配命中后默认拉 yesterday 09:00 +0800 → today 09:00 +0800 的窗口,事件按 PRs merged / created / closed-without-merge / 还开着 + Issues created / closed / 还开着 六桶分组,再按主题二次归类,全文用 呷呷~ 鸭鸭 口吻包裹。
03 SKILL.md 正文
____ _ ___ _ __ __ ___ _____ _ _ _ ___ _ _ ____
| _ \ / \ |_ _| | \ \ / / / __|_ _|/ \ | \ | | \| | | | _ \
| | | |/ _ \ | || | \ V / \__ \ | | / _ \ | \| | |) | | | | |_) |
| |_| / ___ \ | || |___| | ___) || |/ ___ \| |\ | __/| |_| | __/
|____/_/ \_\___|_____|_| |____/ |_/_/ \_\_| \_|____/ \___/|_|
TeamBrain 24h delta → PRs/Issues 六桶分组 → 主题归类 → 鸭语汇报
┌─────────────┐ ┌──────────────┐ ┌──────────────┐
gh ─→ 时间窗过滤 ─→ 6 桶事件分组 ─→ 主题归类 ─→ 鸭语 + still-open 待办
└─────────────┘ └──────────────┘ └──────────────┘
daily-standup — TeamBrain 24h delta report
本 skill 是 TeamBrain 专用。它依赖 libz-renlab-ai/TeamBrain 的 issue/PR 命名公约(fixedflow / issue-N 子任务)和 label 公约(grill-ready / docs-grill-ready / grill-working / ready-for-human / epic),用于让维护者在每天早上一眼看清「昨天上班到今早上班之间,repo 上发生了什么事」。
与邻居 skill 的边界
| 用户在问什么 | 跑哪个 skill |
|---|---|
| 「过去 24h 发生了什么」「since yesterday」 | 本 skill (daily-standup) ← 时间窗 delta |
| 「所有 open issue」「issue 全景」 | repo-issues-status ← 当前快照 |
「今日开工晨报」「/newday」 |
newday ← worktree + open work,不限 24h |
| 「#330 怎么样」 | 裸 gh issue view 330 / gh pr view 330 |
触发条件
只要用户说出下列任一意图(中英都算),立即跑这个 skill,不要先口头说「让我看看」就裸跑 gh pr list:
what happened since yesterday/what happened since yesterday 9am/since yesterday morningexplore today's issues and PRs/today's issues and PRsdaily standup//daily-standup/daily reportshow me yesterday's activity/yesterday → today/last 24 hours on GitHub今天的 issue 和 PR/昨天到今天发生了啥/过去 24 小时 GitHub 活动9 点到 9 点发生了什么/北京 9 点到北京 9 点
如果用户没指定时间窗,默认 = 昨天 09:00 北京时间 → 今天 09:00 北京时间。如果用户给了别的 cutoff("过去 12h"、"上周一到今天"、"since Monday morning"),按用户的时间窗算。
执行步骤
1. 算出 UTC 窗口
北京时间 = UTC+8。默认 cutoff = 09:00 +0800 = 01:00 UTC。所以默认窗口:
[yesterday 01:00 UTC, today 01:00 UTC)
先跑 date -u 拿当前 UTC,确认实际窗口端点:
date -u # 当前 UTC,决定 today
date -u -v-1d +%Y-%m-%dT01:00:00Z 2>/dev/null \
|| date -u -d 'yesterday' +%Y-%m-%dT01:00:00Z # yesterday 01:00 UTC
macOS date -v 和 GNU date -d 两种语法都要试一下,跨平台兼容。
把两个端点存到 shell 变量:
WIN_START="2026-05-12T01:00:00Z" # 示例值,按当前日期算
WIN_END="2026-05-13T01:00:00Z"
2. 拉 PR 和 issue 原始 JSON(两次 gh 调用并行)
gh pr list / gh issue list 的 --search "updated:..." 接受日期粒度(YYYY-MM-DD..YYYY-MM-DD),所以要把窗口稍微放宽一天(前后各 +1 day),然后用 jq 在客户端精确过滤到 UTC 端点。
# PRs (state=all 包含 OPEN + CLOSED + MERGED)
gh pr list --repo libz-renlab-ai/TeamBrain \
--state all --limit 150 \
--search "updated:2026-05-11..2026-05-14" \
--json number,title,state,author,createdAt,updatedAt,closedAt,mergedAt,isDraft,labels \
> /tmp/teamagent/daily-standup-prs.json
# Issues (gh search prs/issues 有 graphql EOF flake,优先 gh issue list)
gh issue list --repo libz-renlab-ai/TeamBrain \
--state all --limit 200 \
--search "updated:2026-05-11..2026-05-14" \
--json number,title,state,author,createdAt,updatedAt,closedAt,labels \
> /tmp/teamagent/daily-standup-issues.json
注意:
- 用
gh pr list --search/gh issue list --search,不要用gh search prs—— 后者会撞 graphql EOF flake。 mergedAt只在gh pr listJSON 字段里有,gh search prs没有,再多一个不能用的理由。- limit 给到 PR=150 / issue=200 是因为 TeamBrain 一天可能 merge 60+ PR(实测 2026-05-12 一天 60 PR),不要给小数字。如果实际返回数等于 limit,加 ⚠️ 标注 「数据可能被截断」。
3. 用 jq 精确过滤 + 六桶分组
复用同一份 jq 模板,把 WIN_START / WIN_END 插进去:
cat /tmp/teamagent/daily-standup-prs.json | jq -r --arg s "$WIN_START" --arg e "$WIN_END" '
def inwin(t): t >= $s and t < $e;
"=== PRs MERGED in window ===",
(map(select(.mergedAt != null and inwin(.mergedAt))) | sort_by(.mergedAt) |
.[] | "[\(.mergedAt | .[5:16])] #\(.number) \(.title) — \(.author.login)"),
"",
"=== PRs CREATED in window ===",
(map(select(inwin(.createdAt))) | sort_by(.createdAt) |
.[] | "[\(.createdAt | .[5:16])] #\(.number) [\(.state)] \(.title) — \(.author.login)"),
"",
"=== PRs CLOSED-without-merge in window ===",
(map(select(.closedAt != null and .mergedAt == null and inwin(.closedAt))) | sort_by(.closedAt) |
.[] | "[\(.closedAt | .[5:16])] #\(.number) \(.title) — \(.author.login)"),
"",
"=== PRs still OPEN as of now ===",
(map(select(.state == "OPEN")) | sort_by(.updatedAt) |
.[] | "[\(.updatedAt | .[5:16])] #\(.number) \(.title) — \(.author.login)")
'
cat /tmp/teamagent/daily-standup-issues.json | jq -r --arg s "$WIN_START" --arg e "$WIN_END" '
def inwin(t): t >= $s and t < $e;
"=== Issues CREATED in window ===",
(map(select(inwin(.createdAt))) | sort_by(.createdAt) |
.[] | "[\(.createdAt | .[5:16])] #\(.number) [\(.state)] \(.title) — \(.author.login)"),
"",
"=== Issues CLOSED in window ===",
(map(select(.closedAt != null and inwin(.closedAt))) | sort_by(.closedAt) |
.[] | "[\(.closedAt | .[5:16])] #\(.number) \(.title) — \(.author.login)"),
"",
"=== Issues still OPEN (updated in window) ===",
(map(select(.state == "OPEN" and inwin(.updatedAt))) | sort_by(.updatedAt) |
.[] | "[\(.updatedAt | .[5:16])] #\(.number) \(.title) — \(.author.login) — labels: \([.labels[].name] | join(\",\"))")
'
六桶 = (PRs merged / PRs created / PRs closed-w/o-merge / PRs still open) + (Issues created / Issues closed / Issues still open in window)。注意 PRs created 列出来的 PR 可能同时出现在 PRs merged 桶(因为它在同一窗口内被创建并合并)—— 这是 feature,不是 bug,方便维护者一眼看 same-day-merge 的速度。
4. 主题归类(汇报时按主题排,不按编号)
读完六桶后,按主题二次归类:
| 主题 | 怎么识别 |
|---|---|
| Feature N(业务特性) | 标题含 feat(m{N}) / feature-N / Feature 1/2/3,参考 docs/BUSINESS-FEATURES.md 三大特性定位 |
| Frontend / Landing | 标题含 landing / RocketTeam / frontend / apps/landing |
| Symphony 第二 driver | 标题含 symphony |
| FIXEDFLOW / driver / grill | 标题含 fixedflow / grill / driver / cross-host mutex |
| m5 propagation | 标题含 m5 / propagation / slice + 编号 |
| 救火 / bug fix | label 含 bug,或标题含 fix(...) 且非 fix(ci) / fix(docs) |
| CI / infra | 标题含 fix(ci) / chore / infra |
| Docs / canned answer | 标题含 docs(...) |
汇报时按上表顺序组织段落,每个主题一段,每段列出 PR/issue 编号 + 一句话翻译(不写实现细节,写「用户能感知的功能」)。
5. 还开着、需要看一眼(最重要的一段)
最后给一张「⚠️ 还开着」表格,4 列:
| PR/Issue | 状态 | 谁的 | 建议下一步 |
只列:
- PR:state == OPEN 且在窗口内更新过的,或在窗口内 created 但 still open
- Issue:state == OPEN 且 (label 含
grill-ready+docs-grill-ready即 driver-dispatchable),或在窗口内被 mention/close-attempted 但还开着的epic/ready-for-human
「建议下一步」要具体,比如:
- 「driver-dispatchable,可直接
/fixed-flow-driver <N>」 - 「等
/reviewPASS 再 squash-merge」 - 「epic,等 child issue 全 close 再人工 close(
ready-for-human不允许 agent close)」
6. 鸭语包裹 + self-report
按 user-level memory feedback_always_duck_voice.md:
- 开场一句
呷呷~ 鸭鸭把过去 24 小时的 GitHub 活动捋完了。窗口 = ... 北京时间。 - 收尾一句
🦆 鸭语小结段,用 ASCII 颜文字(>ω<)/呷呷~ - 结尾给
result:一行 self-contained 完成信号(per session bg-job contract)
最后挂 12-field <self-report> block(project Stop hook 强制要求)。
不要做的事
- 不要裸
gh pr list然后口头描述 —— 必须走 jq 客户端过滤,时间端点对齐到 UTC 1-minute 粒度。 - 不要用
gh search prs—— graphql EOF flake,且没mergedAt字段。 - 不要漏掉「PRs created」桶 —— 用户想看 24h 内多少新工作流入,不只是多少合并。
- 不要把 limit 给得太小 —— TeamBrain 单日 PR 流量可破 60,PR=150 / Issue=200 是 floor。
- 不要忽略
closedAt && !mergedAt的 PR —— 那些是 superseded / abandoned,对维护者复盘很关键。 - 不要把「还开着」段省掉 —— 这是 standup 报告最重要的一段,决定维护者今天先点哪个。
- 不要自动跑
git pull//fixed-flow-driver/ 任何写操作 —— 本 skill 是只读报告 skill,不能改 repo 状态。
04 同仓库更多 Skills
zoom-out
Tell the agent to zoom out and give broader context or a higher-level perspective. Use when you're unfamiliar with a section of code or need to understand how it fits into the bigger picture.
triage
Triage issues through a state machine driven by triage roles. Use when user wants to create an issue, triage issues, review incoming bugs or feature requests, prepare issues for an AFK agent, or manage issue workflow.
to-prd
Turn the current conversation context into a PRD and publish it to the project issue tracker. Use when user wants to create a PRD from the current context.
to-issues
Break a plan, spec, or PRD into independently-grabbable issues on the project issue tracker using tracer-bullet vertical slices. Use when user wants to convert a plan into issues, create implementation tickets, or break down work into issues.
reverification
By-hand only — DO NOT auto-trigger. When the user types /reverification (or asks for "reverify this", "independent verification", "LLM-uncheatable check", "第三方再验证", "用 repo 之外的方法验证"), re-verify recently-completed work using verification methods that **existed before this repo existed** and that an LLM agent cannot fabricate. Pick 3-5 verifications from a 10-family menu (hardware / cryptographic / blockchain / external real-system / mathematical / temporal / content authenticity / physical / network / third-party trust), dispatch them via real tools that capture raw evidence, and report PASS / FAIL / INCONCLUSIVE based ONLY on captured evidence — never on internal reasoning. Refuse to fall back on TeamBrain's bespoke verifiers (judge harness, fastprobe, /review, feature-verification gates) because those did NOT exist before the repo.
repo-issues-status
TeamBrain 仓库**全量 issue 全景**报告(all-scope only)。仅当用户问"status of all issues" / "show me all issues" / "what issues are open" / "all open issues" / "issue 状态(全景)" / "所有 issue 状态" / "issue 全景" / "仓库 issue 概览" / "还有哪些 open issue" 这类**仓库级、多 issue、概览类**问题时触发。**不要在以下情况触发**:(a) 用户问单个 issue 的状态("what's the status of issue