onboard – skill Agent & Tự động hóa cho agent AI
/cs-onboard — Phỏng vấn founder để điền ~/.claude/company-context.md theo schema chuẩn 7 chiều cs-onboard. Lệnh đầu tiên cần chạy khi bắt đầu với c-level-agents. Dùng khi thiết lập bộ C-suite ảo cho công ty mới, hoặc khi cố vấn thiếu bối cảnh công ty — ví dụ trước /c-level-agent đầu tiên
Cài skill onboard bằng một lệnh
Chọn công cụ bạn đang dùng, copy lệnh và dán vào terminal. Skill sẽ nằm trong thư mục skills của công cụ đó và sẵn sàng dùng ở phiên làm việc tiếp theo.
Claude Code
mkdir -p ~/.claude/skills && curl -fsSL https://skill.dn1nguoi.com/z/onboard-c36bb.tgz | tar xz -C ~/.claude/skills
Gọi trong Claude Code: /onboard – rồi mô tả công việc bạn muốn làm.
Codex
mkdir -p ~/.codex/skills && curl -fsSL https://skill.dn1nguoi.com/z/onboard-c36bb.tgz | tar xz -C ~/.codex/skills
Gọi trong Codex: $onboard – rồi mô tả công việc bạn muốn làm.
Hermes Agent
mkdir -p ~/.hermes/skills && curl -fsSL https://skill.dn1nguoi.com/z/onboard-c36bb.tgz | tar xz -C ~/.hermes/skills
Gọi trong Hermes Agent: /onboard – rồi mô tả công việc bạn muốn làm.
DeepSeek
mkdir -p ~/.deepcode/skills && curl -fsSL https://skill.dn1nguoi.com/z/onboard-c36bb.tgz | tar xz -C ~/.deepcode/skills
Gọi trong DeepSeek: /onboard – rồi mô tả công việc bạn muốn làm.
Nội dung SKILL.md
Nội dung SKILL.md gốc bằng tiếng Anh.
/cs-onboard — Founder Interview
Command: /cs-onboard
The first command to run when adopting c-level-agents. A structured founder interview that produces ~/.claude/company-context.md — the file every cs-* advisor reads before responding. Without this, the advisors are guessing.
What This Produces
~/.claude/company-context.md — a single file with the durable facts about the company. Read by:
cs-chief-of-staff(routing decisions)- Every cs-* advisor (context for any question)
/c-level-agents:brief(assumptions in any new decision)
The Interview (12 Questions)
Company Basics
1. Company name and one-sentence pitch.
2. Stage: pre-seed / seed / Series A / Series B / Series C+ / public
3. Headcount: total, by function (eng / product / GTM / ops / G&A)
4. Geographic distribution: HQ + remote split, key countries
Business Model
5. Revenue model: SaaS subscription / usage / transaction / marketplace / hardware / services
6. ICP: name one real customer and describe what they have in common with others
7. ACV: median and range; deal count last 12 months
8. Growth rate: ARR YoY; if pre-revenue, leading metric (users, MAU, etc.)
Financial Posture
9. Runway: months of cash at current burn; bear-case months
10. Last raise: amount, valuation, lead investor, date
Strategic Context
11. Top 3 priorities for the current quarter (in plain language)
12. Top 3 risks the founder loses sleep over (be specific)
Output Format
Canonical schema: ~/.claude/company-context.md is owned by the [cs-onboard](${CLAUDE_PLUGIN_ROOT}/vendor/c-level-advisor/cs-onboard/SKILL.md) skill and follows its 7-dimension schema (${CLAUDE_PLUGIN_ROOT}/vendor/c-level-advisor/cs-onboard/templates/company-context-template.md): Company Identity, Stage & Scale, Founder Profile, Team & Culture, Market & Competition, Current Challenges, Goals & Ambition. The 12 questions above are a faster structured intake that populates that same file — Identity/Business/Financial → Stage & Scale, Team → Team & Culture, Quarter priorities/risks → Current Challenges + Goals & Ambition. Write [not captured] for dimensions the quick intake doesn't reach (Founder Profile, Market & Competition); run the full cs-onboard interview to fill them. Never create a second context file or a divergent layout.
The intake summary captured by the 12 questions:
# Company Context
**Generated:** YYYY-MM-DD
**Last updated:** YYYY-MM-DD
## Identity
- **Company:** <name>
- **Pitch:** <one sentence>
- **Stage:** <stage>
- **HQ + remote:** <distribution>
## Business
- **Model:** <type>
- **ICP:** <description + named customer>
- **ACV:** $<median> (range $<low> - $<high>)
- **Deal count (LTM):** N
- **ARR growth (YoY):** X%
## Financial
- **Cash on hand:** $<amount>
- **Net burn (monthly):** $<amount>
- **Runway base:** N months
- **Runway bear:** N months
- **Last raise:** $<amount> at $<post> in <month YYYY>, led by <investor>
## Team
- **Total headcount:** N
- **Eng:** N | Product: N | GTM: N | Ops: N | G&A: N
## Quarter
- **Top priorities (Q<X> YYYY):**
1. <priority>
2. <priority>
3. <priority>
- **Top risks:**
1. <risk>
2. <risk>
3. <risk>
## Routing Hints
[Optional: any role the founder wants to use sparingly or rely on heavily]Workflow
1. Walk the founder through all 12 questions
2. Quote founder's own words wherever possible (don't paraphrase the ICP)
3. Save to ~/.claude/company-context.md
4. (Optional) If llm-wiki bridge is configured: symlink to vault
ln -sf ~/company-vault/00-meta/company-context.md ~/.claude/company-context.md
5. Confirm with founder: read the file back, ask "anything missing?"
When to Re-Run
- After a fundraise (numbers change)
- After a major pivot or product launch
- After 6+ months (most facts have drifted)
- After a major hire (team distribution changes)
- Always before a
/c-level-agents:boardroomfor a high-stakes decision
Persistence
By default, ~/.claude/company-context.md is local to the founder's machine. To make it persistent across machines / shareable:
- Markdown vault (recommended): see [
../../references/llm-wiki-bridge.md](../../references/llm-wiki-bridge.md) - Encrypted dotfile sync: age + git
- Shared team: keep in a private repo, symlink from
~/.claude/
Related
- Skill: [
cs-onboard](${CLAUDE_PLUGIN_ROOT}/vendor/c-level-advisor/cs-onboard/SKILL.md) — the underlying interview protocol - Skill: [
context-engine](${CLAUDE_PLUGIN_ROOT}/vendor/c-level-advisor/context-engine/SKILL.md) — reads this file - Reference: [
../../references/llm-wiki-bridge.md](../../references/llm-wiki-bridge.md)
---
Version: 1.0.0
Skill liên quan
- design-orchestration – Điều phối quy trình thiết kế bằng cách định tuyến công việc qua brainstorming, review đa…
- stable-baselines3 – Huấn luyện và đánh giá học tăng cường đơn tác nhân với Stable Baselines3 (PPO, SAC, DQN…
- wp-plugin-automation – [Không dịch: mô tả công cụ kiểm thử CVE plugin WordPress hàng loạt trên nhiều domain]
- discord-automation – Tự động hóa tác vụ Discord qua Rube MCP (Composio): tin nhắn, kênh, vai trò, webhook…
- dispatching-parallel-agents – Dùng khi có từ 2 tác vụ trở lên độc lập, có thể làm song song mà không chia sẻ trạng…
- superself – Dùng khi dự án lưu trạng thái trong Superself (khối `<!-- superself:begin` trong…
- distribute-skill-to-all-agents – Phân phối một skill tới các thư mục skill của agent đã cấu hình, đồng thời tôn trọng bố…
- task-coordination-strategies – Phân rã tác vụ phức tạp, thiết kế đồ thị phụ thuộc và điều phối công việc đa agent với…