REA
Reverse engineer anything from your terminal or agent with one CLI and MCP server.
A symbiotic AI agent that remembers everything, challenges you, and extends your cognition.
如果你是作者,可把这段代码贴到 README。徽章会链接回本页。
[](https://mcp.popnx.com/en/skills/lout33-symbiotic-ai)在同一工作中把 AI 连到服务的 MCP
Reverse engineer anything from your terminal or agent with one CLI and MCP server.
Codebase intelligence for TypeScript and JavaScript, exposed as structured MCP analysis tools.
MCP server for terminal commands, file operations, and process management
同类工作,排序与 Skill 列表相同
| 本项 symbiotic-ai | rea | andrej-karpathy-skills | graphify | MoneyPrinterTurbo | |
|---|---|---|---|---|---|
| 作者 | lout33 | morluto | multica-ai | Graphify-Labs | harry0703 |
| Skill 数 | 1 | 1 | 1 | 1 | 1 |
| GitHub 星标 | 738 | 468,733 | 234,332 | 150,287 | 144,566 |
| 最近代码更新 | 2026年6月27日 | 2026年11月19日 | 2026年4月20日 | 2026年11月18日 | 2026年11月18日 |
| 许可证 | – | MIT | – | Apache-2.0 | MIT |
lout33 提供的 1 个 Skill(指令集),AI 读取后照做 · 不是 MCP server
MCP = 手
把 AI 连到外部服务或数据,例如读取 Notion 或开发票。需要有程序或服务器在运行。
Skill = 手册
教 AI 怎么做某项工作的文本文件,不连接任何服务。
有些会重叠:少数 Skill 带有调用工具的脚本,少数 MCP 只提供知识。请从你想完成的工作出发。
Claude CodeOpenCode
依据开发者 README 和 repo 中的文件夹
GitHub account lout33 (person), 8 years old
★ 738 stars · code last changed 2026年6月27日
No, plain text files the AI reads and follows
No scripts, so it does not go online by itself
No hooks or writes to machine settings found
Checked automatically from the repo files at commit ec9df97 (read only, nothing was run). Not a safety guarantee: read it and decide for yourself
第一步:你用哪个 AI?
选择你用的应用,本页只显示该应用的安装方法
This skill has no license, so we do not hand out a ZIP from our site. Download it from GitHub and arrange the folder yourself, or use "Let the AI install it" above, which is easier
Download from GitHub (whole repo) ↗
Unzip it, then zip only the skills/symbiotic-onboard folder again. The ZIP must have a symbiotic-onboard folder with SKILL.md at the top level, no loose files
+Create skillUpload a skillpick symbiotic-onboard.zip
SettingsCapabilitiesCode execution and file creation
Turn on Code execution and file creation first (newer menus: Create and edit files) · personal plans: Settings → Capabilities · Team/Enterprise: an admin turns it on under Organization settings → Plugins & skills · every plan, including Free
Upload once; it works in claude.ai and Claude Desktop, and in recent Claude Code signed in with a Claude account (not with an API key)
Cowork and cloud sessions use the skills turned on in your claude.ai account, not ~/.claude/skills on your computer (Claude Desktop scheduled tasks run locally and still load it)
Checked against official docs 2026年11月12日 · Source: support.claude.com, support.claude.com, code.claude.com
复制这句话,粘贴到 Claude Code 的对话中
For one project, paste it into a chat that is open in that project’s folder · The AI always shows you the path before installing
The AI tells you what it will do first and asks your permission at every step
重新打开 Claude Code,输入 /symbiotic-onboard,或直接提出相关任务,Claude 会自动选用
Checked against official docs 2026年11月12日 · Source: code.claude.com
Skill 是 AI 会照做的指令。安装前先阅读文件,只安装可信开发者的 Skill。
Mac:按 Command+空格输入 Terminal · Windows:下方命令在 PowerShell 中无法使用,请改用上方的“让 AI 帮你安装”
这条命令把 repo 下载到临时文件夹,只复制含 SKILL.md 的文件夹
正在检查命令 · 通过后才会显示复制按钮
ChatGPT 桌面应用内置 Codex,Codex 从 ~/.agents/skills 文件夹读取 Skill
在 ChatGPT 桌面应用中打开 Codex,把这句话粘贴到对话中
The AI always shows you the path before installing
The AI tells you what it will do first and asks your permission at every step
把 symbiotic-onboard 文件夹放到 ~/.agents/skills/symbiotic-onboard
~ 是你的主文件夹($HOME)
Skill 是 AI 会照做的指令。安装前先阅读文件,只安装可信开发者的 Skill。
Mac:按 Command+空格输入 Terminal · Windows:下方命令在 PowerShell 中无法使用,请改用上方的“让 AI 帮你安装”
这条命令把 repo 下载到临时文件夹,只复制含 SKILL.md 的文件夹
正在检查命令 · 通过后才会显示复制按钮
Checked against official docs 2026年11月15日 · Source: learn.chatgpt.com
不能使用你自己安装的 Skill,只能使用目录中插件自带的 Skill
想在 ChatGPT 中使用这个 Skill,请按上面的方法使用 ChatGPT 桌面应用
Checked against official docs 2026年11月15日 · Source: learn.chatgpt.com
复制这句话,粘贴到 Codex 的对话中 · 适用于 ChatGPT 桌面应用中的 Codex、Codex CLI 和 IDE 扩展
For one project, paste it into a chat that is open in that project’s folder · The AI always shows you the path before installing
The AI tells you what it will do first and asks your permission at every step
Codex 会自动发现新安装的 Skill;如果没有出现,请重启 Codex · 使用方法:在 ChatGPT 中输入 @ 并选择 Skill · 在 Codex CLI 或 IDE 扩展中运行 /skills,或输入 $ 加 Skill 名称
Checked against official docs 2026年11月15日 · Source: learn.chatgpt.com
Skill 是 AI 会照做的指令。安装前先阅读文件,只安装可信开发者的 Skill。
Mac:按 Command+空格输入 Terminal · Windows:下方命令在 PowerShell 中无法使用,请改用上方的“让 AI 帮你安装”
这条命令把 repo 下载到临时文件夹,只复制含 SKILL.md 的文件夹
正在检查命令 · 通过后才会显示复制按钮
复制这句话,粘贴到 Cursor 的对话中
For one project, paste it into a chat that is open in that project’s folder · The AI always shows you the path before installing
The AI tells you what it will do first and asks your permission at every step
在 Cursor 的 Agent 聊天中输入 / 并选择 symbiotic-onboard
Checked against official docs 2026年11月12日 · Source: cursor.com
Skill 是 AI 会照做的指令。安装前先阅读文件,只安装可信开发者的 Skill。
Mac:按 Command+空格输入 Terminal · Windows:下方命令在 PowerShell 中无法使用,请改用上方的“让 AI 帮你安装”
这条命令把 repo 下载到临时文件夹,只复制含 SKILL.md 的文件夹
正在检查命令 · 通过后才会显示复制按钮
适用于其他能读取 Skill 的应用。这句话列出了 Claude Code、Codex 和 Cursor 的路径,AI 会按你的应用选择 · Copy this sentence into the app’s chat
For one project, paste it into a chat that is open in that project’s folder · The AI always shows you the path before installing
The AI tells you what it will do first and asks your permission at every step
Checked against official docs 2026年11月12日 · Source: code.claude.com, learn.chatgpt.com, cursor.com
其他支持 Agent Skills 的应用(如 Gemini CLI、GitHub Copilot)使用不同文件夹,请查看该应用的指南,再把含 SKILL.md 的文件夹复制过去。
已于 2026年11月11日 对照 Claude Code、Codex 和 Cursor 官方指南核实 skill 文件夹
以下文字来自开发者的 repo,不是 PopNx MCP 的指令
显示前 8,000 个字符 · 在 GitHub 打开完整文件
---
name: symbiotic-onboard
description: Onboard a new user into a Symbiotic AI setup by interviewing them one sharp question at a time and writing their own context files (SOUL.md, USER.md, NOW.md, and AGENTS.md) as plain Markdown they keep. Use this whenever someone is setting up a personal AI agent, wants their assistant to remember or understand them across sessions, mentions Symbiotic AI / SOUL.md / USER.md / NOW.md, is staring at blank template files full of [bracketed placeholders], or asks how to give an AI persistent context they own and can move between models. Trigger even if they do not say the word "onboard".
metadata:
author: symbiotic-ai
version: 1.0.0
---
# Symbiotic Onboard
## What this does and why
This skill sets a brand-new user up with a working Symbiotic AI configuration in one short conversation. The whole point of Symbiotic AI is **ownership and portability**: a person's context lives in plain Markdown files they own, can read and edit by hand, can version, and can carry to any model. Native platform memory cannot offer that. It is opaque, automatic, and locked to one vendor.
So the job here is not to "store" anything in a hidden place. It is to **interview the user and write plain files they keep.** When you finish, the user has four files (`SOUL.md`, `USER.md`, `NOW.md`, and optionally `AGENTS.md`) that any agent runtime can load. They own them.
The hard part of onboarding is the blank page. A new user opens template files full of `[bracketed placeholders]`, does not know what to type, and stalls. This skill removes that wall by doing a short, sharp interview and filling the files for them.
## The core rule
**Interview one sharp question at a time. Never dump a list of questions.** A wall of questions makes people freeze or write thin, generic answers. One question, wait, write the answer into the right file, then ask the next. This is how you get real content instead of a survey.
Stay in **SOUL voice** while you do it: short, direct, references what they just said, no filler. You are a co-pilot who is going to know this person, not an intake form.
## Setup before the interview
1. Decide where the files go. Default to the current working directory. If the user has a workspace or repo for their agent, use that. Ask in one line if unclear: "Where should your files live? (default: here)"
2. Seed the files from the bundled templates so the structure is correct and the user can see what a finished setup looks like:
- Copy `assets/SOUL.md`, `assets/USER.md`, `assets/NOW.md`, and `assets/AGENTS.md` into the target directory.
- `SOUL.md` and `AGENTS.md` are already filled (agent identity and operating protocol). You usually leave them as-is. Offer to tune `SOUL.md` voice later if the user wants a different personality.
- `USER.md` and `NOW.md` contain `[bracketed placeholders]`. These are what the interview fills.
3. Tell the user, briefly, what is about to happen and why it matters: "I am going to ask you a few sharp questions and write your answers into plain files you own. A handful of questions now, the rest fills in as we work together."
If the user would rather start from a realistic example than a blank template, that is fine: copy a filled example profile over the templates and rewrite it as them, one answer at a time. Editing a real example beats filling blanks. (In the Symbiotic AI repo these live under `examples/`.)
## The interview
Ask these first. They are enough to make the agent useful today. Ask, wait, write, move on.
1. **Name** - "What should I call you?" -> write into `USER.md` Identity.
2. **Current goal** - "What are you actually trying to make happen right now?" -> `USER.md` Mission / The Real Goal, and reflect the near-term version into `NOW.md`.
3. **What they are avoiding** - "What is the thing you keep not doing?" -> `USER.md` Psychology (start a "Bugs" entry) and/or Who You Are.
4. **Today's MIT** - "If you finished one thing today and felt good tonight, what is it?" -> `NOW.md` under `# QUEUE > ## TODAY` as the first checkbox.
Then, only if the user has energy for more, you can deepen: stack/skills, what charges vs drains them, how they work best, a recent win. Each still one question at a time, each written into the matching `USER.md` section.
**Defer the heavy stuff.** Financial reality and deep psychology (drivers, fears, long patterns) are real sections in `USER.md`, but do not mine for them in session one. Leave those placeholders for later sessions. Pushing for money and trauma on day one kills the onboarding. Get them operational first; the deep profile accumulates over time, which is exactly the Symbiotic advantage.
### How to write the answers
- Replace the matching `[bracketed placeholder]`, do not append next to it. The file should read like the user wrote it.
- Keep their voice. If they say "I keep rewriting my landing page instead of shipping," write that, not "Tendency toward perfectionism."
- Mirror anything raw before moving on. If an answer is heavy, acknowledge it in one line, then continue. This is setup, not therapy, but do not steamroll a real moment.
- After each write, you can briefly confirm what you captured so they see the file taking shape and trust that it is theirs.
## Finish the file: defer what you did not ask
This step matters as much as the interview. A short interview fills the high-value fields (name, goal, avoidance, today's MIT) but leaves dozens of secondary `[bracketed placeholders]` in `USER.md` and `NOW.md` (age, location, stack, energy map, relationships, financials). If you leave those raw, a first-time user sees broken-looking brackets and assumes the skill half-failed. Do not turn the interview into a 40-question wall to avoid this. Handle it at the end instead.
As your final write, sweep `USER.md` and `NOW.md` and resolve every remaining `[bracketed placeholder]` you did not fill, using whichever reads cleaner per field:
- **Replace it with a deferred marker** so the structure stays visible but reads as intentional: `_(not set yet, your agent will fill this in as it learns you)_`. Use this for fields worth keeping in view (stack, energy map, psychology bugs, milestones).
- **Drop the whole optional subsection** when an empty scaffold adds noise rather than value (for example a `Career` or `Shipped` table the user has nothing for yet). Removing it is cleaner than a table of deferred markers.
Leave markdown checkboxes like `[ ]` alone. Those are list syntax, not placeholders. You are only resolving the descriptive `[Your ...]`-style fill-ins.
The end state is firm: **no raw `[Your ...]`-style brackets remain.** Every field is either filled in the user's words or shows the clear deferred treatment. The file should read as complete and intentional, never as broken output.
Because the deferred marker uses parentheses, not brackets, it will not re-trigger the agent's onboarding protocol on the next session. The bundled `AGENTS.md` tells the agent to fill these deferred fields naturally as it learns the user, so they complete through use rather than a wall of questions.
## Close
When name, goal, avoidance, and today's MIT are written, the deferred sweep is done, and no raw `[bracketed placeholders]` remain:
1. Confirm: "You are set up. I know who you are and what today is about."
2. State today's MIT back to them in their own words.
3. Tell them how to actually use the files (see below).
4. One line forward: "Go."
## Tell them how to use the files
This is the step that makes the work real. End by showing the user how their new files plug into an agent. Keep it to what fits their setup:
- **The files are yours.** They are plain Markdown in their directory. They can open, edit, correct, delete, and version them in git anytime. Nothing is hidden.
- **At session start, the agent reads all four** (`SOUL.md`, `USER.md`, `AGENTS.md`, then `NOW.md`) and acts from them. `NOW.md` updates as they work; the stable files change rarely提问、分享技巧或报告无法使用。按“有帮助”数量排序。