К основному содержанию
скилл Claude Code CX Codex Cursor GitHub Copilot официальный · Anthropic

stakeholder-update — скилл для Claude Code: отчёты для команды

Скилл готовит обновления о прогрессе под аудиторию и формат: от статуса для руководства до сообщения клиентам. Пригодится для регулярного отчёта, запуска или эскалации риска.

Установить в Claude Code · Через Skills CLI (Vercel) Все способы и платформы ↓
npx skills add anthropics/knowledge-work-plugins --skill stakeholder-update -a claude-code
Открыть на GitHub

Что делает stakeholder-update

Скилл помогает собрать обновление из прогресса, препятствий, решений и следующих шагов. Можно задать еженедельный или ежемесячный формат, сообщить о запуске либо описать отдельную ситуацию. Для руководства он предлагает краткую сводку с рисками и решениями, для инженеров — технический контекст и блокеры, для клиентов — пользу и известные проблемы без внутреннего жаргона.

Если подключены проектный трекер, чат, расшифровки встреч или база знаний, скилл ищет там статусы, обсуждения и решения. Без подключённых инструментов он попросит перечислить сделанное, текущие риски и ближайшие планы. Это повод заранее проверить, какие источники доступны: без них контекст придётся дать самостоятельно.

После черновика можно попросить изменить тон или уровень подробностей и подготовить формат для письма, сообщения, документа или слайдов. Для клиентского текста предусмотрено описание известных проблем и обходных решений, если они есть. Исходное описание обрывается на разделе о статусной отчётности, поэтому здесь приведены только раскрытые сценарии.

Какие задачи решает

  • Собрать еженедельный отчёт о прогрессе, блокерах и следующих шагах
  • Подготовить сводку для руководства с рисками и нужными решениями
  • Объяснить инженерам технические блокеры и изменения приоритетов
  • Сообщить клиентам о запуске и известных проблемах
  • Переложить один статус в версии для разных аудиторий

Примеры запросов агенту

  • Подготовь еженедельный статус для руководства: прогресс, риски и решения, которые от меня нужны.
  • Составь инженерное обновление по запуску: что готово, что блокирует работу и что дальше.
  • Перепиши этот статус для клиентов: расскажи о пользе запуска и известных проблемах без внутреннего жаргона.

Как установить stakeholder-update

1Через Skills CLI (Vercel)

npx skills add anthropics/knowledge-work-plugins --skill stakeholder-update -a claude-code

2Вручную: положить файлы

~/.claude/skills/stakeholder-update/SKILL.md

Папка скилла целиком; для одного проекта — .claude/skills/stakeholder-update/

Из того же набора89

Весь набор →
data-visualization официальный anthropics Подбирает подходящие типы графиков под структуру данных и генерирует код на Python с matplotlib, seaborn и plotly. Пригодится при подготовке отчётов. Claude Code CX Codex Cursor +1 13 700установок documentation официальный anthropics Скилл для технического письма: собирает README, справочник по API, ранбук, архитектурный документ или гайд для новичков. Пригодится, когда код есть, а описания к нему нет. Claude Code CX Codex Cursor +1 10 300установок build-dashboard официальный anthropics Генерирует автономный HTML-дашборд с графиками Chart.js, карточками KPI и таблицами. Помогает быстро оформить результаты запроса или сырые данные в один файл для браузера. Claude Code CX Codex Cursor +1 9 100установок memory-management официальный anthropics Скилл раскладывает рабочие сокращения, имена и термины по файлам памяти, чтобы агент учитывал их в запросах. Пригодится для проектов с внутренним словарём и множеством контекстных деталей. Claude Code CX Codex Cursor +1 7 400установок task-management официальный anthropics Ведёт список задач в локальном файле TASKS.md и показывает интерактивный дашборд. Пригодится, чтобы фиксировать дедлайны и поручения прямо во время диалога с агентом. Claude Code CX Codex Cursor +1 7 200установок architecture официальный anthropics Помогает оформить Architecture Decision Record или разобрать чужой дизайн системы: варианты, компромиссы, последствия. Пригодится, когда выбираете между технологиями. Claude Code CX Codex Cursor +1 6 700установок knowledge-synthesis официальный anthropics Скилл объединяет результаты корпоративного поиска из чатов, почты и документов в один связный ответ. Убирает дубли, выставляет уровень доверия и сохраняет ссылки на источники. Claude Code CX Codex Cursor +1 6 400установок testing-strategy официальный anthropics Скилл помогает спроектировать стратегию тестирования и план проверок для проекта. Пригодится, когда нужно выбрать типы тестов и найти пробелы в покрытии. Claude Code CX Codex Cursor +1 6 400установок

Похожие

code-documentation bytedance Скилл автоматически создаёт README, API-референс, архитектурную документацию и гайды разработчика прямо из кода. Пригодится, если нужно документировать проект, библиотеку или API без ручной писанины. Claude Code CX Codex Cursor +1 1 900установок changelog-automation Seth Hobson Скилл помогает собирать changelog и release notes из коммитов, PR и релизов. Подойдёт для настройки процесса релизов и единых правил коммитов. Claude Code CX Codex Cursor 13 733установок golang-swagger samber Помогает добавлять и поддерживать OpenAPI/Swagger-документацию в Go-проектах через аннотации и генерацию схемы. Пригодится для описания API и Swagger UI. Claude Code CX Codex Cursor +1 37 000установок adev-writing-guide angular Скилл задаёт правила технического писательства Google и разметку для документации Angular в adev/. Он нужен при создании, правке и ревью страниц в каталоге adev/src/content. Claude Code CX Codex Cursor +1 912установок firecrawl-download firecrawl Скилл выкачивает веб-сайт или его разделы в локальные файлы markdown и скриншоты. Помогает подготовить офлайн-документацию для работы агента. Claude Code CX Codex Cursor +1 79 800установок grill-with-docs Matt Pocock Скилл проводит жёсткое интервью по плану или дизайну и по ходу создаёт ADR и глоссарий. Подойдёт перед началом сложной разработки. Claude Code 1 219 756установок skill-creator официальный anthropics Помогает создать новый скилл, доработать существующий и проверить результат через тестовые промпты и оценки. Claude Code Claude.ai 414 180установок setup-matt-pocock-skills Matt Pocock Скилл настраивает репозиторий для инженерных скиллов Matt Pocock: задаёт трекер задач, метки триажа и расположение доменной документации. Запустите его перед первым использованием остальных скиллов. Claude Code 1 052 851установок

Вопросы о stakeholder-update

Как установить stakeholder-update в Claude Code?

Выполните команду npx skills add anthropics/knowledge-work-plugins --skill stakeholder-update -a claude-code.

Что делает stakeholder-update?

Он готовит обновления о статусе проекта для выбранной аудитории и формата: например, для руководства, инженеров или клиентов. Источниками контекста могут быть подключённые инструменты или сведения от пользователя.

Какие источники контекста использует stakeholder-update?

При подключении он может искать сведения в проектном трекере, чате, расшифровках встреч и базе знаний. Если инструменты не подключены, он попросит пользователя описать прогресс, риски, решения и планы.

Исходный текст

SKILL.md · Apache-2.0
---
name: stakeholder-update
description: Generate a stakeholder update tailored to audience and cadence. Use when writing a weekly or monthly status for leadership, announcing a launch, escalating a risk or blocker, or translating the same progress into exec-brief, engineering-detail, or customer-facing versions.
argument-hint: "<update type and audience>"
---

# Stakeholder Update

> If you see unfamiliar placeholders or need to check which tools are connected, see [CONNECTORS.md](../../CONNECTORS.md).

Generate a stakeholder update tailored to the audience and cadence.

## Usage

```
/stakeholder-update $ARGUMENTS
```

## Workflow

### 1. Determine Update Type

Ask the user what kind of update:
- **Weekly**: Regular cadence update on progress, blockers, and next steps
- **Monthly**: Higher-level summary with trends, milestones, and strategic alignment
- **Launch**: Announcement of a feature or product launch with details and impact
- **Ad-hoc**: One-off update for a specific situation (escalation, pivot, major decision)

### 2. Determine Audience

Ask who the update is for:
- **Executives / leadership**: High-level, outcome-focused, strategic framing, brief
- **Engineering team**: Technical detail, implementation context, blockers, decisions needed
- **Cross-functional partners**: Context-appropriate detail, focus on shared goals and dependencies
- **Customers / external**: Benefits-focused, clear timelines, no internal jargon
- **Board**: Metrics-driven, strategic, risk-focused, very concise

### 3. Pull Context from Connected Tools

If **~~project tracker** is connected:
- Pull status of roadmap items and milestones
- Identify completed items since last update
- Surface items that are at risk or blocked
- Pull sprint or iteration progress

If **~~chat** is connected:
- Search for relevant team discussions and decisions
- Find blockers or issues raised in channels
- Identify key decisions made asynchronously

If **~~meeting transcription** is connected:
- Pull recent meeting notes and discussion summaries
- Find decisions and action items from relevant meetings

If **~~knowledge base** is connected:
- Search for recent meeting notes
- Find decision documents or design reviews

If no tools are connected, ask the user to provide:
- What was accomplished since the last update
- Current blockers or risks
- Key decisions made or needed
- What is coming next

### 4. Generate the Update

Structure the update for the target audience using the templates and frameworks below.

**For executives**: TL;DR, status color (G/Y/R), key progress tied to goals, decisions made, risks with mitigation, specific asks, and next milestones. Keep it under 300 words.

**For engineering**: What shipped (with links), what is in progress (with owners), blockers, decisions needed (with options and recommendation), and what is coming next.

**For cross-functional partners**: What is coming that affects them, what you need from them (with deadlines), decisions that impact their team, and areas open for input.

**For customers**: What is new (framed as benefits), what is coming soon, known issues with workarounds, and how to provide feedback. No internal jargon.

**For launch announcements**: What launched, why it matters, key details (scope, availability, limitations), success metrics, rollout plan, and feedback channels.

### 5. Review and Deliver

After generating the update:
- Ask if the user wants to adjust tone, detail level, or emphasis
- Offer to format for the delivery channel (email, chat post, doc, slides)
- If **~~chat** is connected, offer to draft the message for sending

## Update Templates by Audience

### Executive / Leadership Update
Executives want: strategic context, progress against goals, risks that need their help, decisions that need their input.

**Format**:
```
Status: [Green / Yellow / Red]

TL;DR: [One sentence — the most important thing to know]

Progress:
- [Outcome achieved, tied to goal/OKR]
- [Milestone reached, with impact]
- [Key metric movement]

Risks:
- [Risk]: [Mitigation plan]. [Ask if needed].

Decisions needed:
- [Decision]: [Options with recommendation]. Need by [date].

Next milestones:
- [Milestone] — [Date]
```

**Tips for executive updates**:
- Lead with the conclusion, not the journey. Executives want "we shipped X and it moved Y metric" not "we had 14 standups and resolved 23 tickets."
- Keep it under 200 words. If they want more, they will ask.
- Status color should reflect YOUR genuine assessment, not what you think they want to hear. Yellow is not a failure — it is good risk management.
- Only include risks you want help with. Do not list risks you are already handling unless they need to know.
- Asks must be specific: "Decision on X by Friday" not "support needed."

### Engineering Team Update
Engineers want: clear priorities, technical context, blockers resolved, decisions that affect their work.

**Format**:
```
Shipped:
- [Feature/fix] — [Link to PR/ticket]. [Impact if notable].

In progress:
- [Item] — [Owner]. [Expected completion]. [Blockers if any].

Decisions:
- [Decision made]: [Rationale]. [Link to ADR if exists].
- [Decision needed]: [Context]. [Options]. [Recommendation].

Priority changes:
- [What changed and why]

Coming up:
- [Next items] — [Context on why these are next]
```

**Tips for engineering updates**:
- Link to specific tickets, PRs, and documents. Engineers want to click through for details.
- When priorities change, explain why. Engineers are more bought in when they understand the reason.
- Be explicit about what is blocking them and what you are doing to unblock it.
- Do not waste their time with information that does not affect their work.

### Cross-Functional Partner Update
Partners (design, marketing, sales, support) want: what is coming that affects them, what they need to prepare for, how to give input.

**Format**:
```
What's coming:
- [Feature/launch] — [Date]. [What this means for your team].

What we need from you:
- [Specific ask] — [Context]. By [date].

Decisions made:
- [Decision] — [How it affects your team].

Open for input:
- [Topic we'd love feedback on] — [How to provide it].
```

### Customer / External Update
Customers want: what is new, what is coming, how it benefits them, how to get started.

**Format**:
```
What's new:
- [Feature] — [Benefit in customer terms]. [How to use it / link].

Coming soon:
- [Feature] — [Expected timing]. [Why it matters to you].

Known issues:
- [Issue] — [Status]. [Workaround if available].

Feedback:
- [How to share feedback or request features]
```

**Tips for customer updates**:
- No internal jargon. No ticket numbers. No technical implementation details.
- Frame everything in terms of what the customer can now DO, not what you built.
- Be honest about timelines but do not overcommit. "Later this quarter" is better than a date you might miss.
- Only mention known issues if they are customer-impacting and you have a resolution plan.

## Status Reporting Framework

### Green / Yellow / Red Status

**Green** (On Track):
- Progressing as planned
- No significant risks or blockers
- On track to meet commitments and deadlines
- Use Green when things are genuinely going well — not as a default

**Yellow** (At Risk):
- Progress is slower than planned, or a risk has materialized
- Mitigation is underway but outcome is uncertain
- May miss commitments without intervention or scope adjustment
- Use Yellow proactively — the earlier you flag risk, the more options you have

**Red** (Off Track):
- Significantly behind plan
- Major blocker or risk without clear mitigation
- Will miss commitments without significant intervention (scope cut, resource addition, timeline extension)
- Use Red when you genuinely need help. Do not wait until it is too late.

### When to Change Status
- Move to Yellow at the FIRST sign of risk, not when you are sure things are bad
- Move to Red when you have exhausted your own options and need escalation
- Move back to Green only when the risk is genuinely resolved, not just paused
- Document what changed when you change status — "Moved to Yellow because [reason]"

## Risk Communication

### ROAM Framework for Risk Management
- **Resolved**: Risk is no longer a concern. Document how it was resolved.
- **Owned**: Risk is acknowledged and someone is actively managing it. State the owner and the mitigation plan.
- **Accepted**: Risk is known but we are choosing to proceed without mitigation. Document the rationale.
- **Mitigated**: Actions have reduced the risk to an acceptable level. Document what was done.

### Communicating Risks Effectively
1. **State the risk clearly**: "There is a risk that [thing] happens because [reason]"
2. **Quantify the impact**: "If this happens, the consequence is [impact]"
3. **State the likelihood**: "This is [likely/possible/unlikely] because [evidence]"
4. **Present the mitigation**: "We are managing this by [actions]"
5. **Make the ask**: "We need [specific help] to further reduce this risk"

### Common Mistakes in Risk Communication
- Burying risks in good news. Lead with risks when they are important.
- Being vague: "There might be some delays" — specify what, how long, and why.
- Presenting risks without mitigations. Every risk should come with a plan.
- Waiting too long. A risk communicated early is a planning input. A risk communicated late is a fire drill.

## Decision Documentation (ADRs)

### Architecture Decision Record Format
Document important decisions for future reference:

```
# [Decision Title]

## Status
[Proposed / Accepted / Deprecated / Superseded by ADR-XXX]

## Context
What is the situation that requires a decision? What forces are at play?

## Decision
What did we decide? State the decision clearly and directly.

## Consequences
What are the implications of this decision?
- Positive consequences
- Negative consequences or tradeoffs accepted
- What this enables or prevents in the future

## Alternatives Considered
What other options were evaluated?
For each: what was it, why was it rejected?
```

### When to Write an ADR
- Strategic product decisions (which market segment to target, which platform to support)
- Significant technical decisions (architecture choices, vendor selection, build vs buy)
- Controversial decisions where people disagreed (document the rationale for future reference)
- Decisions that constrain future options (choosing a technology, signing a partnership)
- Decisions you expect people to question later (capture the context while it is fresh)

### Tips for Decision Documentation
- Write ADRs close to when the decision is made, not weeks later
- Include who was involved in the decision and who made the final call
- Document the context generously — future readers will not have today's context
- It is okay to document decisions that were wrong in hindsight — add a "superseded by" link
- Keep them short. One page is better than five.

## Meeting Facilitation

### Stand-up / Daily Sync
**Purpose**: Surface blockers, coordinate work, maintain momentum.
**Format**: Each person shares:
- What they accomplished since last sync
- What they are working on next
- What is blocking them

**Facilitation tips**:
- Keep it to 15 minutes. If discussions emerge, take them offline.
- Focus on blockers — this is the highest-value part of standup
- Track blockers and follow up on resolution
- Cancel standup if there is nothing to sync on. Respect people's time.

### Sprint / Iteration Planning
**Purpose**: Commit to work for the next sprint. Align on priorities and scope.
**Format**:
1. Review: what shipped last sprint, what carried over, what was cut
2. Priorities: what are the most important things to accomplish this sprint
3. Capacity: how much can the team take on (account for PTO, on-call, meetings)
4. Commitment: select items from the backlog that fit capacity and priorities
5. Dependencies: flag any cross-team or external dependencies

**Facilitation tips**:
- Come with a proposed priority order. Do not ask the team to prioritize from scratch.
- Push back on overcommitment. It is better to commit to less and deliver reliably.
- Ensure every item has a clear owner and clear acceptance criteria.
- Flag items that are underscoped or have hidden complexity.

### Retrospective
**Purpose**: Reflect on what went well, what did not, and what to change.
**Format**:
1. Set the stage: remind the team of the goal and create psychological safety
2. Gather data: what went well, what did not go well, what was confusing
3. Generate insights: identify patterns and root causes
4. Decide actions: pick 1-3 specific improvements to try next sprint
5. Close: thank people for honest feedback

**Facilitation tips**:
- Create psychological safety. People must feel safe to be honest.
- Focus on systems and processes, not individuals.
- Limit to 1-3 action items. More than that and nothing changes.
- Follow up on previous retro action items. If you never follow up, people stop engaging.
- Vary the retro format occasionally to prevent staleness.

### Stakeholder Review / Demo
**Purpose**: Show progress, gather feedback, build alignment.
**Format**:
1. Context: remind stakeholders of the goal and what they saw last time
2. Demo: show what was built. Use real product, not slides.
3. Metrics: share any early data or feedback
4. Feedback: structured time for questions and input
5. Next steps: what is coming next and when the next review will be

**Facilitation tips**:
- Demo the real product whenever possible. Slides are not demos.
- Frame feedback collection: "What feedback do you have on X?" is better than "Any thoughts?"
- Capture feedback visibly and commit to addressing it (or explaining why not)
- Set expectations about what kind of feedback is actionable at this stage

## Output Format

Keep updates scannable. Use bold for key points, bullets for lists. Executive updates should be under 300 words. Engineering updates can be longer but should still be structured for skimming.

## Tips

- The most common mistake in stakeholder updates is burying the lead. Start with the most important thing.
- Status colors (Green/Yellow/Red) should reflect reality, not optimism. Yellow is not a failure — it is good risk communication.
- Asks should be specific and actionable. "We need help" is not an ask. "We need a decision on X by Friday" is.
- For executives, frame everything in terms of outcomes and goals, not activities and tasks.
- If there is bad news, lead with it. Do not hide it after good news.
- Match the length to the audience's attention. Executives get a few bullets. Engineering gets the details they need.

© anthropics · Apache-2.0 · anthropics/knowledge-work-plugins

Смотрите также

Спонсоры проекта

Компании, которые помогают держать аналитику открытой