skills/aaaaqwq__agi-super-team
1
0
Fork 0
mirror of https://github.com/aAAaqwq/AGI-Super-Team.git synced 2026-09-14 16:33:39 +08:00
Code Issues Packages Projects Releases Wiki Activity

docs(brand): rebuild the evidence-first project story

Browse Source
This commit is contained in:
aAAaqwq
2026-07-21 04:07:55 -07:00
parent c05a108d94
commit b0771b46a2
21 changed files with 474 additions and 697 deletions
Download Patch File Download Diff File Expand all files Collapse all files
.github/ISSUE_TEMPLATE/bug_report.md
+26
View File
@@ -0,0 +1,26 @@
---
name: Bug report
about: Report a reproducible repository problem
title: "[Bug]: "
labels: bug
---
Do not disclose vulnerabilities or secrets here. Use the private channel in [SECURITY.md](../../SECURITY.md).
## Revision and environment
- Commit:
- OS:
- Harness and version:
## Reproduction
Provide minimal, sanitized steps and inputs.
## Expected and actual behavior
Describe both, including relevant sanitized logs.
## Verification and recovery
What checks have you run? Is there a workaround or rollback?
.github/ISSUE_TEMPLATE/feature_request.md
+26
View File
@@ -0,0 +1,26 @@
---
name: Feature request
about: Propose an evidence-backed improvement
title: "[Feature]: "
labels: enhancement
---
## Problem and user
Who has the problem, and what evidence shows it matters?
## Proposed outcome
Describe the smallest useful outcome, not only an implementation.
## Alternatives and boundaries
What alternatives were considered? What must remain manual or require explicit authorization?
## Evidence and verification
How would maintainers test value, safety, compatibility, and failure recovery?
## Provenance
List relevant sources, licenses, and prior art.
.github/PULL_REQUEST_TEMPLATE.md
+30
View File
@@ -0,0 +1,30 @@
## Summary
What changes, and why?
## Scope and risk
- Affected paths:
- User-visible behavior before/after:
- Security, privacy, financial, publishing, or destructive-operation risks:
- Rollback or removal plan:
## Evidence
- [ ] `npm test`
- [ ] `npm run validate`
- [ ] Examples were run in an isolated environment where safe
- [ ] Skipped checks and reasons are listed below
Commands, results, fixtures, harness versions, and limitations:
## Provenance
For copied or adapted material, list source URL, immutable revision or retrieval date, license, and adaptation. Write `Original` if none applies.
## Safety and review
- [ ] No secrets, private data, session material, or production configuration
- [ ] No unsupported catalog, compatibility, performance, profit, security, or production-readiness claims
- [ ] External actions require explicit human approval
- [ ] Documentation and local links were updated
CODE_OF_CONDUCT.md
+17
View File
@@ -0,0 +1,17 @@
# Code of Conduct
## Our standard
We welcome constructive participation regardless of background or experience. Be respectful, specific, and evidence-oriented. Critique ideas and changes, not people.
Unacceptable behavior includes harassment, threats, discriminatory language, unwanted sexual attention, doxxing, deliberate spam, impersonation, and publishing another person's private information without permission.
## Project spaces
This standard applies to repository issues, pull requests, discussions, reviews, and project-related public representation. Marketing and community participation must follow platform rules and the ethical guidance in [`growth/`](./growth/README.md).
## Enforcement
Report conduct concerns privately to project maintainers through an available private repository channel. Do not use the security advisory channel unless the report also concerns a security vulnerability.
Maintainers may edit or remove content, limit participation, or ban contributors based on the context and severity of behavior. They should handle reports discreetly, avoid conflicts of interest, and protect reporters from retaliation.
CONTRIBUTING.md
+31 -62
View File
@@ -1,83 +1,52 @@
# Contributing to AGI Super Team
Thank you for your interest in contributing! This guide covers how to add skills, agents, and improvements.
Contributions should be reviewable, attributable, safe to test, and honest about support. Do not include secrets, private data, unlicensed content, or claims that cannot be reproduced.
> **Harness-agnostic policy:** AGI Super Team works natively with **Claude Code, Codex, Cursor, and Hermes**. Skills and agents must stay harness-neutral — never hard-depend on a single tool's CLI (e.g. the discontinued `openclaw` CLI). Reference harness capabilities generically; if a skill needs a specific tool, document it as optional and provide an alternative path.
## Before opening a pull request
## Adding a Skill
1. Branch from `main` and keep the change focused.
2. Search for an existing skill, agent, or issue before adding a duplicate.
3. Record provenance for adapted material: upstream URL, revision or retrieval date, license, and what changed.
4. Add or update tests for executable behavior. For documentation, run examples in an isolated temporary destination when safe.
5. Run the repository checks:
### Directory Structure
```
skills/
└── your-skill-name/
└── SKILL.md # Required: skill definition
└── scripts/ # Optional: helper scripts
└── assets/ # Optional: images, templates
```
```bash
npm test
npm run validate
```
### SKILL.md Format
```markdown
# Skill Name
6. Review the diff for credentials, personal data, unsafe shell commands, generated artifacts, and unsupported claims.
One-line description of what this skill does.
## Skills
## When to Use
- Trigger condition 1
- Trigger condition 2
A skill lives at `skills/<name>/SKILL.md`. Keep one clear purpose, document triggers and boundaries, and place optional scripts or assets in the same skill directory.
## Instructions
Step-by-step guidance for the AI agent.
Commands must state prerequisites, expected effects, and recovery. Destructive, external, production, or account-changing operations need explicit human confirmation. Never embed credentials or encourage users to pipe unreviewed remote code into a shell.
## Examples
Concrete usage examples.
```
## Agents
### Guidelines
- **One skill = one purpose** — keep it focused
- **Concrete, not vague** — include specific commands, file paths, parameters
- **Language**: English or Chinese, pick one per skill (don't mix)
- **No credentials**: never include API keys, tokens, or passwords
- **No binary files > 1MB**: use links to external assets instead
- **Test locally**: verify the skill works with your AI coding harness (Claude Code, Codex, Cursor, Hermes) before submitting
An agent lives at `agents/<id>/` and normally includes `SOUL.md` and `AGENTS.md`; other persona or workflow files are optional. Keep role boundaries clear and use mentor references only as creative framing, not affiliation or endorsement.
## Adding an Agent
## Provenance and licensing
### Directory Structure
```
agents/
└── your-agent-id/
├── SOUL.md # Required: personality, mentor, philosophy
├── AGENTS.md # Required: role, responsibilities, collaboration
└── TOOLS.md # Optional: skill index with ../skills/ links
```
For copied or adapted content, include the source project, immutable revision where possible, source license, and adaptation summary in the relevant provenance file or pull-request description.
### Guidelines
- Each agent should have a distinct **spirit mentor** that shapes their thinking style
- Include a 32-thinking framework skill if possible
- Reference skills via relative paths (`../skills/skill-name/`)
- Keep SOUL.md concise — personality, not documentation
Confirm that the source license permits redistribution. Do not submit scraped private content, proprietary prompts, model outputs with unclear rights, or generated code copied from an unverified source.
## Reporting Issues
## Testing evidence
When reporting bugs, include:
1. What you expected to happen
2. What actually happened
3. Your AI harness and model you're using
4. Any relevant logs or error messages
The pull request must include:
## PR Process
- exact commands executed and their results;
- the harness and relevant version used for manual checks;
- fixtures or sanitized inputs needed to reproduce the result;
- limitations, skipped checks, and untested environments;
- screenshots only when they add evidence and contain no sensitive data.
1. Fork the repository
2. Create a feature branch: `git checkout -b feature/your-feature`
3. Commit with clear messages: `git commit -m "add: new skill for X"`
4. Push and open a PR
5. Respond to review feedback
Do not claim “production-ready,” “live-validated,” profitable, secure, or universally compatible solely from unit tests or a local demo.
## Code of Conduct
## Pull request scope
- Be respectful and constructive
- Focus on the technical merit of contributions
- No personal attacks or political content in skills
- Keep skills vendor-neutral where possible
Use `.github/PULL_REQUEST_TEMPLATE.md`. Describe behavior before and after, risk, rollback, provenance, and evidence. Maintainers may ask for a smaller change or reject additions that cannot be safely maintained.
Thank you for making this project better! 🚀
Report vulnerabilities privately as described in [SECURITY.md](./SECURITY.md). Follow [CODE_OF_CONDUCT.md](./CODE_OF_CONDUCT.md) in all project spaces.
README.md
+93 -249
View File
@@ -1,279 +1,123 @@
<p align="center">
<img src="assets/logo.png" alt="AGI Super Team" width="120">
<img src="assets/banner.png" alt="AGI Super Team — Evidence-backed AI teams for real outcomes" width="100%">
</p>
<h1 align="center">AGI Super Team</h1>
<p align="center">
<strong>Deploy a virtual C-Suite of AI executives — each powered by a legendary mind.</strong><br/>
Let Jim Simons run your quant · MrBeast drive your content · Buffett guard your capital.
</p>
<p align="center">
<img src="https://img.shields.io/badge/Works%20with-Claude%20Code%20%7C%20Codex%20%7C%20Cursor-blueviolet" alt="Harness compatible">
<a href="./LICENSE"><img src="https://img.shields.io/badge/License-MIT-green.svg" alt="MIT License"></a>
</p>
<p align="center"><strong>Evidence-backed AI teams for real outcomes</strong></p>
<p align="center">
<a href="./README_CN.md">中文</a> ·
<a href="#quick-start">Quick Start</a> ·
<a href="./skills/README.md">All Skills</a> ·
<a href="./agents/README.md">Agents</a> ·
<a href="./cookbook/">Cookbooks</a> ·
<a href="./starter-kits/">Starter Kits</a> ·
<a href="./setup.md">Agent Setup Guide</a>
<a href="#safe-quick-start">Quick start</a> ·
<a href="./setup.md">Setup guide</a> ·
<a href="./CONTRIBUTING.md">Contribute</a>
</p>
---
## What this repository is
## 🤖 For AI Agents
AGI Super Team is a collection of agent personas, reusable skills, starter-kit manifests, and a curated Codex-native package. It helps people assemble reviewable AI workflows rather than promising autonomous business results.
> **If you are Claude Code / Codex / Cursor (or any compatible AI harness), read this file: [`setup.md`](./setup.md)**
>
> Or just tell your Agent:
> ```
> Install AGI Super Team for me: https://github.com/aAAaqwq/AGI-Super-Team
> I want the solo-founder kit
> ```
Outputs still need human review. Trading, legal, security, medical, publishing, and deployment tasks require domain-specific validation and appropriate authorization.
## 💡 What Is This?
## Choose a distribution
A **plug-and-play AI team template** — deploy a complete virtual C-Suite using Claude Code, Cursor, Codex, or Gemini. Each agent is shaped by a **spirit mentor** (Elon Musk, Jensen Huang, Warren Buffett, Richard Feynman...) that defines their personality, decision-making style, and cognitive framework.
| Surface | Repository support | Install path | Notes |
|---|---|---|---|
| Generic/local workspace | Supported by `install.sh` | Preview, inspect, then apply | Copies selected personas and available skills without overwriting existing files |
| Codex | Separate curated native package | See [`.codex/INDEX.md`](./.codex/INDEX.md) | Has its own manifest, selected skills, and opt-in agent sync |
| Claude Code | Repository plugin manifest present | Review [`.claude-plugin/`](./.claude-plugin/) | Confirm your installed client supports the manifest before use |
| Cursor, Gemini, Kimi | Metadata/manifests present | Review the corresponding manifest | Compatibility varies by client version; this repository does not claim feature parity |
**1,639 skills. 14 agents. 31 thinking frameworks. 30 workflows.** Zero boilerplate — copy, customize, ship.
The generic starter kits and Codex package are separate products. Installing one does not install or synchronize the other.
## ⚡ Install
## Safe quick start
> **Works natively with Claude Code, Cursor, Codex, and Gemini.**
Clone a trusted revision and preview the local installer. Preview is read-only and is the default.
```bash
# Claude Code (recommended)
/plugin install aAAaqwq/AGI-Super-Team
# OpenAI Codex — curated native swarms, memory, and 31 specialist agents
codex plugin marketplace add aAAaqwq/AGI-Super-Team --ref main
codex plugin add agi-super-team-codex@agi-super-team
# Or clone directly
git clone --depth 1 https://github.com/aAAaqwq/AGI-Super-Team.git ~/.agi-super-team
```
See the [Codex package index](./.codex/INDEX.md) for the agent catalog, safe synchronization workflow, update commands, and provenance.
For one-command kit deploys (solo-founder, quant-trader, content-creator…), jump to [What You Can Do in 10 Minutes](#what-you-can-do-in-10-minutes).
## 🚀 What You Can Do in 10 Minutes
Three killer use cases — each backed by battle-tested skills that ship in the box, bound to the right legendary mind.
### 📈 Quant Trading — Jim Simons in your corner
Deploy the **quant-trader kit** and the CQO agent (Jim Simons) immediately runs production strategies:
- **`5minbtc`** — BTC 5-minute direction prediction (v5.7.3 engine, live-validated)
- **`a-share-analysis`** — A-share quant analysis with a one-click daily script
- **`a-fund-monitor`** — real-time fund NAV estimation + Telegram alerts
```bash
curl -sSL https://raw.githubusercontent.com/aAAaqwq/AGI-Super-Team/main/install.sh | bash -s -- quant-trader
```
### 📣 Viral Content — MrBeast's playbook
Deploy the **content-creator kit** and the CCO agent (MrBeast) takes over:
- **`xhs-content-creator`** / **`xhs-skill`** — Xiaohongshu (RED) content built for the algorithm
- **`wechat-article-writer`** / **`wechat-ai-radar`** — WeChat MP articles + morning briefings
- **`content-cover-gen`** — content-driven cover image generation
```bash
curl -sSL https://raw.githubusercontent.com/aAAaqwq/AGI-Super-Team/main/install.sh | bash -s -- content-creator
```
### 🏛️ Full C-Suite — your AI-native company
All 14 executives, 1,639 skills, 31 thinking frameworks — one command:
```bash
curl -sSL https://raw.githubusercontent.com/aAAaqwq/AGI-Super-Team/main/install.sh | bash -s -- full-team
```
> Each kit drops the right skills into `~/.openclaw/workspace-{agent}/skills/` and binds them to the matching C-Suite agent. No manual wiring.
## 🏛️ Architecture
```
You (Founder / Chairman)
└── 👑 CEO — Strategy, coordination, quality gate
├── ⚡ CTO — Architecture, technology strategy, R&D direction
├── 🎨 CPO — Product design, UX, brand DNA
├── 📈 CQO — Quant trading, algorithmic strategies
├── 📣 CMO — Marketing, SEO, growth
├── 💰 CFO — Finance, P&L, capital allocation
├── 📊 CDO — Data pipelines, analytics, governance
├── ✍️ CCO — Content creation, viral growth
├── ⚖️ CLO — Legal, compliance, IP protection
├── 🔬 CRO — Deep research, frontier intelligence
├── 🤝 CSO — Sales, BD, revenue growth
├── ⚙️ COO — Ops, OKRs, cross-team coordination
├── 💻 PE — Production engineering, DevOps, delivery
└── ⚖️ Governor — Quality assurance, delivery audit, escalation
```
<p align="center">
<img src="https://img.shields.io/badge/Skills-1,639-blueviolet" alt="Skills">
<img src="https://img.shields.io/badge/Agents-14-orange" alt="Agents">
<img src="https://img.shields.io/badge/Frameworks-31-cyan" alt="Frameworks">
</p>
## 👥 Agents
| Agent | Role | Spirit Mentor | Thinking |
|-------|------|---------------|----------|
| [`ceo`](./agents/ceo/) | 👑 CEO | Elon Musk | First Principles, Cross-Domain Synthesis |
| [`cto`](./agents/cto/) | ⚡ CTO | Jensen Huang | Systems Thinking, Accelerated Computing |
| [`cpo`](./agents/cpo/) | 🎨 CPO | Steve Jobs | Design Thinking, Radical Simplicity |
| [`cqo`](./agents/cqo/) | 📈 CQO | Jim Simons | Mathematical Rigor, Probabilistic Thinking |
| [`cmo`](./agents/cmo/) | 📣 CMO | David Ogilvy | Data-Driven Storytelling, Audience Psychology |
| [`cfo`](./agents/cfo/) | 💰 CFO | Warren Buffett | Value Investing, Margin of Safety |
| [`cdo`](./agents/cdo/) | 📊 CDO | Nate Silver | Bayesian Reasoning, Predictive Analytics |
| [`cco`](./agents/cco/) | ✍️ CCO | MrBeast | Viral Mechanics, Platform Algorithms |
| [`clo`](./agents/clo/) | ⚖️ CLO | Alan Dershowitz | Legal Reasoning, Risk Assessment |
| [`cro`](./agents/cro/) | 🔬 CRO | Richard Feynman | Scientific Method, Feynman Technique |
| [`cso`](./agents/cso/) | 🤝 CSO | Michael Dell | Solution Selling, Relationship Building |
| [`coo`](./agents/coo/) | ⚙️ COO | Andy Grove | High Output Management, OKR Discipline |
| [`pe`](./agents/pe/) | 💻 PE | Linus Torvalds, antirez, DHH | Pragmatic Engineering, Ship Over Talk |
| [`governor`](./agents/governor/) | ⚖️ Governor | Zhuge Liang, Wang Yangming | Tri-Verification, Evidence-Based Audit |
> Each agent folder contains `SOUL.md` (personality), `AGENTS.md` (operations), `TOOLS.md` (skill links). Fully customizable.
## 🧠 Thinking Frameworks
31 distilled thinking skills based on real-world mentors — mental models, decision frameworks, and classic quotes with sources:
```bash
# Inject a mentor's thinking into any agent
cp -r skills/thinking-elon-musk/ ~/.openclaw/workspace-ceo/skills/
# Or inject all frameworks to all workspaces
for agent in ceo cto cpo cqo cmo cfo cdo cco clo cro cso coo pe; do
mkdir -p ~/.openclaw/workspace-${agent}/skills/
cp -r skills/thinking-* ~/.openclaw/workspace-${agent}/skills/
done
```
## 🛠️ Skill Categories
| Category | Highlights |
|----------|-----------|
| 🔌 SaaS Integrations | Notion, Airtable, HubSpot, Stripe, ActiveCampaign, 60+ more |
| 📝 Content & Writing | SEO, viral copy, anti-AI-slop, social media |
| 🔧 Development | Backend, frontend, Docker, Git, TDD, API design, code review |
| 💰 Trading & Finance | Crypto, Polymarket, DeFi, portfolio management, backtesting |
| 🤖 AI Agent Patterns | Multi-agent orchestration, parallel execution, sub-agents |
| 📊 Data & Analytics | Web scraping, DuckDB, CSV pipelines, arXiv |
| 📈 Marketing & SEO | SEO audits, GEO optimization, A/B testing, competitor analysis |
| 🎨 Design & Media | Image generation, UI/UX, brand identity |
| 🏢 Business & Strategy | SaaS launch, competitor teardown, financial modeling |
| 📋 Project Management | PRD, roadmaps, Scrum, team coordination |
| 💬 Communication | Email, Feishu, WeChat, Telegram, LinkedIn, cross-instance messaging |
| 📱 Chinese Platforms | Xiaohongshu, Douyin, WeChat MP, Juejin, Zhihu |
| ⚙️ DevOps & Infra | AWS, Docker, Linux, observability, deployment |
| 🧬 Bioinformatics | Genome analysis, metagenomics, pharmacogenomics |
| 🎬 Video & Digital Human | Video editing, digital human, storyboard, subtitle |
| 🤖 Web3 & Autonomys | Decentralized storage, auto-deploy, auto-memory |
👉 **[Skill-Agent Mapping Matrix →](./SKILL-AGENT-MATRIX.md)** — Each skill mapped to its C-Suite agent (PE, CTO, CQO, etc.)
👉 **[Full skill catalog →](./skills/README.md)**
## 🔄 Workflows
30 production-ready workflows across all C-Suite roles:
| Scope | Examples |
|-------|---------|
| **Shared** | Daily standup, weekly review, crisis escalation, cross-agent handoff |
| **Per-Agent** | Content pipeline (CCO), market morning brief (CQO), P&L tracking (CFO), code review (PE), incident response (COO) |
Each agent directory (e.g. `agents/cco/WORKFLOW.md`) contains role-specific and shared workflows.
## ⚡ Quick Start
For the native plugin install, see [Install](#install) above. For one-command kit deploys, see [What You Can Do in 10 Minutes](#what-you-can-do-in-10-minutes).
### Starter Kits
| Kit | Agents | Best for |
|-----|--------|----------|
| 🚀 [**Solo Founder**](./starter-kits/solo-founder/) | CEO + PE + CCO | Indie hackers, solo founders |
| 🎨 [**Content Creator**](./starter-kits/content-creator/) | CCO + CDO + CMO | Content teams, self-media |
| 📈 [**Quant Trader**](./starter-kits/quant-trader/) | CQO + CDO + CFO | Quant trading, investment |
| 🏛️ **Full Team** | All 14 agents | Complete AI-native company |
### Manual deploy
```bash
git clone https://github.com/aAAaqwq/AGI-Super-Team.git
git clone --depth 1 --branch main https://github.com/aAAaqwq/AGI-Super-Team.git
cd AGI-Super-Team
./install.sh solo-founder # or any kit / agent name
./install.sh --source "$PWD" --destination /path/to/review-workspace solo-founder
```
## 📚 Cookbooks
Review the proposed agents, source, destination, and relevant files. Apply only after the preview matches your intent:
In-depth learning guides in [`cookbook/`](./cookbook/):
| Book | Description |
|------|-------------|
| [Self-Media Operations](./cookbook/self-media-operations-handbook/) | Complete handbook: XHS, Douyin, WeChat, content strategy |
| [Quantitative Trading](./cookbook/quant-learning/) | Crypto trading, algorithmic strategies, risk management |
| [Prompt Engineering](./cookbook/prompt-engineering-learning/) | Advanced prompt techniques and patterns |
| [Knowledge Base](./cookbook/knowledge-book/) | Cross-domain knowledge distillation |
| [Crypto Deep Dive](./cookbook/crypto-learning/) | Blockchain fundamentals and DeFi |
## 📁 Repository Structure
```
AGI-Super-Team/
├── agents/ # 14 C-Suite agent personas
│ ├── ceo/ # SOUL.md · AGENTS.md · TOOLS.md · WORKFLOW.md
│ ├── cto/ # ...
│ ├── governor/ # Quality assurance & delivery audit
│ └── README.md # Architecture diagram & skill matrix
├── skills/ # 2,659 skills (flat structure, each with SKILL.md)
│ └── README.md # Full catalog
├── starter-kits/ # One-click deployment bundles
│ ├── solo-founder/ # CEO + PE + CCO
│ ├── content-creator/ # CCO + CDO + CMO
│ └── quant-trader/ # CQO + CDO + CFO
├── cookbook/ # 5 in-depth learning guides
├── install.sh # One-click deployer
├── CHARTER.md # Team constitution (12 principles)
├── STARTUP.md # Quick-start guide
├── COLLABORATION.md # Inter-agent collaboration network
```bash
./install.sh --source "$PWD" --destination /path/to/review-workspace --apply solo-founder
```
## 🤝 Contributing
The installer preserves existing persona files and skill directories. See [setup.md](./setup.md) for prerequisites, verification, updates, and recovery.
Contributions welcome! See [CONTRIBUTING.md](./CONTRIBUTING.md) for guidelines.
Avoid piping a remote script directly into a shell when you can inspect a pinned checkout first.
1. Fork the repo
2. Create your skill: `skills/your-skill/SKILL.md`
3. Submit a PR
## 📄 License
[MIT](./LICENSE) — use freely, attribution appreciated.
---
## ⭐ Star History
<a href="https://star-history.com/#aAAaqwq/AGI-Super-Team&Date">
<picture>
<source media="(prefers-color-scheme: dark)" srcset="https://api.star-history.com/svg?repos=aAAaqwq/AGI-Super-Team&type=Date&theme=dark" />
<source media="(prefers-color-scheme: light)" srcset="https://api.star-history.com/svg?repos=aAAaqwq/AGI-Super-Team&type=Date" />
<img alt="Star History Chart" src="https://api.star-history.com/svg?repos=aAAaqwq/AGI-Super-Team&type=Date" />
</picture>
</a>
---
### Preview → apply → verify
<p align="center">
Compatible with Claude Code, Codex, Cursor, and Hermes
<img src="assets/demo-install.gif" alt="Terminal demo showing a read-only preview, explicit apply, and repository checks passing" width="760">
</p>
The animation uses sanitized project-relative paths. Read the [static transcript](./assets/demo-install.txt) or run the same commands in a clean checkout.
## Packages and manifests
- [Codex package index](./.codex/INDEX.md): curated Codex skills, specialist roles, sync behavior, provenance, and update policy.
- [Codex marketplace manifest](./.agents/plugins/marketplace.json): points to the separate `plugins/agi-super-team-codex` package.
- [Starter kits](./starter-kits/): small selections for a solo founder, content team, or quantitative research workflow.
- [Agents](./agents/): persona and operating files for the generic workspace installer.
- [Skills](./skills/): the cross-harness skill library. Its contents change; use the repository validator instead of a hard-coded count.
## Starter kits
| Kit | Agents | Intended use |
|---|---|---|
| [Solo Founder](./starter-kits/solo-founder/) | CEO, PE, CCO | Planning, engineering, and reviewed content drafts |
| [Content Creator](./starter-kits/content-creator/) | CCO, CDO, CMO | Research, content drafts, and measurement plans |
| [Quant Trader](./starter-kits/quant-trader/) | CQO, CDO, CFO | Research, backtesting, and risk review—not live trading |
| `full-team` | All repository agents | Broad evaluation; start smaller when possible |
Examples describe tasks the agents can assist with, not validated performance claims. External publishing, financial transactions, and production changes remain manual and human-authorized.
## Evidence and verification
Repository checks are the source of truth for catalog integrity:
```bash
npm test
npm run validate
```
Do not publish exact catalog counts until the validator is green for the revision being described. When making a claim, link to reproducible inputs, record the revision, and distinguish tests from real-world validation.
## Architecture
```text
Founder / operator
└── CEO — coordination and quality gates
├── CTO / PE — architecture and implementation
├── CPO / CCO / CMO — product, content, and growth
├── CQO / CFO / CDO — quantitative research, finance, and data
├── CLO / CRO / CSO / COO — legal, research, sales, and operations
└── Governor — independent review and escalation
```
Each agent directory may contain persona, identity, workflow, and tool guidance. Treat mentor names as creative framing, not affiliation, endorsement, or an imitation guarantee.
## Safety boundaries
- Never place credentials, private user data, browser sessions, or production configuration in a skill or issue.
- Review third-party commands and dependencies before execution.
- Run financial workflows in research or paper-trading environments until independently validated; no bundled strategy is guaranteed profitable or production-ready.
- Require explicit human approval for posts, messages, transactions, deployments, and destructive operations.
- Report vulnerabilities privately through [GitHub Security Advisories](https://github.com/aAAaqwq/AGI-Super-Team/security/advisories/new), not a public issue.
## Project links
- [Setup and recovery](./setup.md)
- [Contributing and provenance](./CONTRIBUTING.md)
- [Security policy](./SECURITY.md)
- [Growth playbooks](./growth/README.md)
- [License](./LICENSE)
## Star History
![Star History](https://aaaaqwq.github.io/AGI-Super-Team/assets/star-history.svg)
README_CN.md
+95 -156
View File
@@ -1,184 +1,123 @@
<p align="center">
<img src="assets/logo.png" alt="AGI Super Team" width="120">
<img src="assets/banner.png" alt="AGI Super Team — 以证据为基础的 AI 团队，服务真实成果" width="100%">
</p>
<h1 align="center">AGI Super Team</h1>
<p align="center">
<strong>部署一支由传奇大脑驱动的 AI 高管团队。</strong><br/>
让 Jim Simons 跑量化 · MrBeast 做爆款 · 巴菲特管钱。
</p>
<p align="center">
<img src="https://img.shields.io/badge/Works%20with-Claude%20Code%20%7C%20Codex%20%7C%20Cursor-blueviolet" alt="Harness compatible">
<a href="./LICENSE"><img src="https://img.shields.io/badge/License-MIT-green.svg" alt="MIT License"></a>
<img src="https://img.shields.io/badge/Skills-1,651-blueviolet" alt="Skills">
<img src="https://img.shields.io/badge/Agents-14-orange" alt="Agents">
<img src="https://img.shields.io/badge/Frameworks-31-cyan" alt="Frameworks">
</p>
<p align="center"><strong>Evidence-backed AI teams for real outcomes</strong><br/>以证据为基础的 AI 团队，服务真实成果</p>
<p align="center">
<a href="./README.md">English</a> ·
<a href="#quick-start">快速开始</a> ·
<a href="./skills/README.md">全部技能</a> ·
<a href="./agents/README.md">Agent</a> ·
<a href="./cookbook/">教程</a> ·
<a href="./workflows/">工作流</a>
<a href="#安全快速开始">快速开始</a> ·
<a href="./setup.md">安装指南</a> ·
<a href="./CONTRIBUTING.md">参与贡献</a>
</p>
---
## 这个仓库是什么
## 💡 这是什么？
AGI Super Team 收录 Agent 人格、可复用技能、starter kit 清单，以及独立维护的 Codex 原生精选包。它帮助用户组装可审查的 AI 工作流，而不是承诺自动取得商业成果。
一个**即插即用的 AI 团队模板** — 兼容 Claude Code / Codex / Cursor 等 AI 编程助手，部署完整的虚拟 C-Suite。每个 Agent 都有**精神导师**（Elon Musk、Jensen Huang、Warren Buffett...）塑造其性格和决策方式。
所有输出仍需人工审查。交易、法律、安全、医疗、发布和部署任务必须经过相应领域验证，并获得明确授权。
**1,651 技能 · 14 个 Agent · 31 个思维框架 · 30 个工作流。** 零配置 — 复制、定制、发布。
## 选择分发方式
## ⚡ 安装
| 使用方式 | 仓库支持 | 安装路径 | 说明 |
|---|---|---|---|
| 通用/本地工作区 | 由 `install.sh` 支持 | 预览、检查、再应用 | 复制所选人格和现有技能，不覆盖已有文件 |
| Codex | 独立的原生精选包 | 见 [`.codex/INDEX.md`](./.codex/INDEX.md) | 使用独立 manifest、精选技能和可选 Agent 同步 |
| Claude Code | 仓库包含 plugin manifest | 检查 [`.claude-plugin/`](./.claude-plugin/) | 使用前确认客户端版本支持该 manifest |
| Cursor、Gemini、Kimi | 仓库包含元数据/manifest | 检查对应 manifest | 各客户端版本能力不同，本仓库不宣称功能完全一致 |
> **原生支持 Claude Code、Cursor、Codex、Gemini。**
通用 starter kit 与 Codex 包是不同的分发物。安装其中一个不会自动安装或同步另一个。
## 安全快速开始
先克隆可信版本，再预览本地安装器。默认预览不会写入目标目录。
```bash
# Claude Code（推荐）
/plugin install aAAaqwq/AGI-Super-Team
# OpenAI Codex —— 精选原生 Swarm、记忆能力与 31 个专业 Agent
codex plugin marketplace add aAAaqwq/AGI-Super-Team --ref main
codex plugin add agi-super-team-codex@agi-super-team
# 或直接克隆
git clone --depth 1 https://github.com/aAAaqwq/AGI-Super-Team.git ~/.agi-super-team
```
完整 Agent 清单、安全同步流程、更新命令与来源说明见 [Codex 包索引](./.codex/INDEX.md)。
一键 kit 部署（solo-founder / quant-trader / content-creator 等）见下方[10 分钟你能做什么](#10-分钟你能做什么)。
## 🚀 10 分钟你能做什么
三个杀手级用例 —— 每个都由仓库内置的实战 skill 支撑，绑定对应的传奇大脑。
### 📈 量化交易 —— Jim Simons 为你操盘
部署 **quant-trader kit**，CQO（Jim Simons）立即开始运行实战策略：
- **`5minbtc`** —— BTC 5 分钟方向预测（v5.7.3 引擎，实盘验证）
- **`a-share-analysis`** —— A 股量化分析，一键每日脚本
- **`a-fund-monitor`** —— 基金净值实时估算 + Telegram 推送
```bash
curl -sSL https://raw.githubusercontent.com/aAAaqwq/AGI-Super-Team/main/install.sh | bash -s -- quant-trader
```
### 📣 爆款内容 —— MrBeast 的玩法
部署 **content-creator kit**，CCO（MrBeast）接管内容：
- **`xhs-content-creator`** / **`xhs-skill`** —— 为小红书算法打造的内容
- **`wechat-article-writer`** / **`wechat-ai-radar`** —— 公众号文章 + 每日情报
- **`content-cover-gen`** —— 内容驱动的封面图生成
```bash
curl -sSL https://raw.githubusercontent.com/aAAaqwq/AGI-Super-Team/main/install.sh | bash -s -- content-creator
```
### 🏛️ 完整高管团队 —— 你的 AI 原生公司
14 位高管、1,651 技能、31 套思维框架 —— 一行命令：
```bash
curl -sSL https://raw.githubusercontent.com/aAAaqwq/AGI-Super-Team/main/install.sh | bash -s -- full-team
```
> 每个 kit 会把对应的 skill 装进 `~/.openclaw/workspace-{agent}/skills/`，并绑定到匹配的高管 Agent。无需手动配置。
## 🏛️ 架构
```
你（创始人 / 董事长）
└── 👑 CEO — 战略、协调、质量把关
├── ⚡ CTO — 代码、架构、调试
├── 🎨 CPO — 产品设计、用户体验、品牌
├── 📈 CQO — 量化交易、市场分析
├── 📣 CMO — 营销、SEO、增长
├── 💰 CFO — 财务、利润、成本优化
├── 📊 CDO — 数据采集、ETL、分析
├── ✍️ CCO — 内容创作、病毒式增长
├── ⚖️ CLO — 法律、合规、合同
├── 🔬 CRO — 深度研究、情报
├── 🤝 CSO — 销售、商务拓展
├── ⚙️ COO — 运营、监控、效率
└── 💻 PE — 全栈工程、DevOps
```
## 🚀 快速开始
> 推荐使用上方[安装](#安装)章节的原生 plugin 入口；下面的手动方式适合需要细粒度控制的用户。
```bash
# 1. 克隆仓库
git clone https://github.com/aAAaqwq/AGI-Super-Team.git
git clone --depth 1 --branch main https://github.com/aAAaqwq/AGI-Super-Team.git
cd AGI-Super-Team
# 2. 部署 Agent（例如 CEO）
mkdir -p ~/.openclaw/workspace-ceo/skills/
cp -r skills/thinking-elon-musk/ ~/.openclaw/workspace-ceo/skills/
# 3. 给任何 Agent 添加技能
cp -r skills/api-design/ ~/.openclaw/workspace-ceo/skills/
# 4. 重启你的 Agent harness — 完成！
./install.sh --source "$PWD" --destination /path/to/review-workspace solo-founder
```
## 🛠️ 技能分类
核对计划安装的 Agent、来源、目标路径和相关文件。确认无误后再显式应用：
| 分类 | 亮点 |
|------|------|
| 🔧 开发 | 前端、后端、Docker、Git、TDD、API 设计 |
| 💰 交易与金融 | 加密货币、Polymarket、量化策略、回测 |
| 📝 内容与写作 | SEO、病毒文案、反AI检测、社交媒体 |
| 📈 营销与SEO | SEO 审计、GEO 优化、竞品分析 |
| 📱 中国平台 | 小红书、抖音、微信公众号、知乎、掘金 |
| 🔌 SaaS 集成 | 60+ 平台：HubSpot、Stripe、Airtable 等 |
| 🎬 视频与媒体 | AI 视频、数字人、FFmpeg、字幕 |
| 🤖 AI Agent 模式 | 多 Agent 编排、并行执行、子 Agent |
| 🧬 生物信息学 | 基因组分析、药物基因组学 |
👉 **[完整技能目录 →](./skills/README.md)**
## 📚 教程
| 教程 | 描述 |
|------|------|
| [自媒体运营手册](./cookbook/self-media-operations-handbook/) | 小红书、抖音、微信完整运营策略 |
| [量化交易](./cookbook/quant-learning/) | 加密货币、算法策略、风险管理 |
| [Prompt 工程](./cookbook/prompt-engineering-learning/) | 高级提示词技术与模式 |
## 📁 仓库结构
```
AGI-Super-Team/
├── agents/ # 14 个 C-Suite Agent
├── skills/ # 1,651 个技能（扁平结构）
├── workflows/ # 30 个标准工作流
├── cookbook/ # 5 个深度教程
├── CHARTER.md # 团队宪章
└── STARTUP.md # 快速开始指南
```bash
./install.sh --source "$PWD" --destination /path/to/review-workspace --apply solo-founder
```
## 📄 许可证
安装器会保留已有的人格文件和技能目录。依赖、验证、更新和恢复方法见 [setup.md](./setup.md)。
[MIT](./LICENSE) — 自由使用，请注明出处。
如能先检查固定版本的 checkout，请不要把远程脚本直接传给 shell 执行。
---
## ⭐ Star History
<a href="https://star-history.com/#aAAaqwq/AGI-Super-Team&Date">
<picture>
<source media="(prefers-color-scheme: dark)" srcset="https://api.star-history.com/svg?repos=aAAaqwq/AGI-Super-Team&type=Date&theme=dark" />
<source media="(prefers-color-scheme: light)" srcset="https://api.star-history.com/svg?repos=aAAaqwq/AGI-Super-Team&type=Date" />
<img alt="Star History Chart" src="https://api.star-history.com/svg?repos=aAAaqwq/AGI-Super-Team&type=Date" />
</picture>
</a>
---
### 预览 → 应用 → 验证
<p align="center">
兼容 Claude Code / Codex / Cursor / Hermes 等 AI 编程助手
<img src="assets/demo-install.gif" alt="终端演示：只读预览、显式应用和仓库检查通过" width="760">
</p>
动画中的路径已经替换为项目内相对路径。你也可以阅读[静态文字记录](./assets/demo-install.txt)，或在干净 checkout 中运行相同命令。
## 包与 Manifests
- [Codex 包索引](./.codex/INDEX.md)：精选 Codex 技能、专业角色、同步行为、来源和更新策略。
- [Codex marketplace manifest](./.agents/plugins/marketplace.json)：指向独立的 `plugins/agi-super-team-codex` 包。
- [Starter kits](./starter-kits/)：独立开发者、内容团队和量化研究的小型组合。
- [Agents](./agents/)：通用工作区安装器使用的人格和操作文件。
- [Skills](./skills/)：跨 harness 技能库。内容会变化，请使用仓库 validator，不要写死数量。
## Starter kits
| Kit | Agents | 适用场景 |
|---|---|---|
| [Solo Founder](./starter-kits/solo-founder/) | CEO、PE、CCO | 规划、工程和经人工审查的内容草稿 |
| [Content Creator](./starter-kits/content-creator/) | CCO、CDO、CMO | 调研、内容草稿和衡量计划 |
| [Quant Trader](./starter-kits/quant-trader/) | CQO、CDO、CFO | 研究、回测和风险审查，不用于实盘交易 |
| `full-team` | 仓库全部 Agents | 广泛评估；通常建议从更小组合开始 |
示例表示 Agent 可以协助的任务，不代表已经验证的效果。对外发布、金融交易和生产变更必须由人工明确授权。
## 证据与验证
仓库完整性以自动检查为准：
```bash
npm test
npm run validate
```
在当前版本 validator 通过之前，不发布精确目录数量。提出成果主张时，应链接可复现输入、记录版本，并明确区分测试结果与真实环境验证。
## 架构
```text
创始人 / 操作者
└── CEO — 协调与质量门禁
├── CTO / PE — 架构与实现
├── CPO / CCO / CMO — 产品、内容与增长
├── CQO / CFO / CDO — 量化研究、财务与数据
├── CLO / CRO / CSO / COO — 法律、研究、销售与运营
└── Governor — 独立审查与升级
```
Agent 目录可能包含人格、身份、工作流和工具指南。导师姓名仅用于创作框架，不表示关联、背书或保证模仿效果。
## 安全边界
- 不要在技能或 issue 中放入凭据、私人数据、浏览器会话或生产配置。
- 执行前审查第三方命令和依赖。
- 金融工作流在独立验证前仅用于研究或模拟交易；仓库不保证任何策略盈利或可直接用于生产。
- 帖子、消息、交易、部署和破坏性操作必须经过人工明确批准。
- 安全漏洞请通过 [GitHub Security Advisories](https://github.com/aAAaqwq/AGI-Super-Team/security/advisories/new) 私密报告，不要提交公开 issue。
## 项目链接
- [安装与恢复](./setup.md)
- [贡献与来源要求](./CONTRIBUTING.md)
- [安全政策](./SECURITY.md)
- [增长手册](./growth/README.md)
- [许可证](./LICENSE)
## Star History
![Star History](https://aaaaqwq.github.io/AGI-Super-Team/assets/star-history.svg)
SECURITY.md
+20 -34
View File
@@ -1,44 +1,30 @@
# Security Policy
# Security policy
## ⚠️ This Repository is De-sensitized
## Report a vulnerability privately
All provider-specific API keys, tokens, secrets, and vendor names have been replaced with generic placeholders.
Use [GitHub Security Advisories](https://github.com/aAAaqwq/AGI-Super-Team/security/advisories/new) to submit a private report. Do not open a public issue, discussion, or pull request for an undisclosed vulnerability.
### Placeholder Conventions
| Pattern | Meaning |
|---------|---------|
| `<provider>/<model>` | Replace with your provider/model ID |
| `your-provider.example.com` | Replace with your API endpoint |
| `YOUR_API_KEY` | Replace with your actual API key |
| `pass show api/<service>` | Reference to local password store |
| `Provider-A`, `Provider-B` | Generic provider names |
Include the affected path and revision, impact, reproduction steps, prerequisites, and a suggested mitigation if available. Remove credentials, personal data, and production identifiers from evidence.
### What is NOT in This Repo
- ❌ Real API keys or tokens
- ❌ `auth-profiles.json` (provider credentials)
- ❌ `models.json` (model routing config)
- ❌ `openclaw.json` (runtime config)
- ❌ `.env` files
- ❌ Private keys (`.key`, `.pem`)
- ❌ Browser session data
If private reporting is unavailable, do not publish exploit details. State only that the private channel is unavailable in a general repository discussion and wait for a maintainer-provided channel.
### Pre-Push Security Protocol
Before every push, run the three-layer scan:
## Scope
```bash
# Layer 1: Critical key patterns
grep -rE 'AIza[A-Za-z0-9_-]{35}|sk-[A-Za-z0-9]{20,}|ghp_[A-Za-z0-9]{36}|xoxb-|AKIA[A-Z0-9]{16}' \
--include='*.py' --include='*.js' --include='*.json' --include='*.md' --include='*.sh' -l .
Reports may cover repository-owned installers, scripts, plugin manifests, bundled skills, agent instructions, or documentation that creates a concrete security risk.
# Layer 2: Broader secret patterns
grep -rE 'Bearer [A-Za-z0-9]{20,}|api[_-]?key.*=.*[A-Za-z0-9]{20,}' \
--include='*.py' --include='*.js' --include='*.json' --include='*.md' --include='*.sh' -l .
Third-party services, harnesses, models, and upstream dependencies are outside this project's control. Report their vulnerabilities to the relevant vendor, while privately notifying this project if its integration also needs mitigation.
# Layer 3: Git history check
git log --all --diff-filter=A --name-only --pretty=format: | sort -u | grep -iE 'secret|token|key|auth|env'
```
## Response expectations
All three layers must pass with zero real credentials before pushing.
Maintainers will aim to acknowledge a report, assess severity, coordinate a fix, and publish remediation guidance. Response times are not guaranteed. Please allow reasonable time for remediation before disclosure.
### Reporting
If you discover a leaked credential in this repo, please open an issue immediately.
## User safety
- Inspect a pinned checkout before running scripts; use the installer's preview before `--apply`.
- Keep API keys, tokens, private keys, browser sessions, and production configuration outside the repository.
- Use least-privilege credentials and isolated test accounts.
- Treat skills and agent output as untrusted instructions until reviewed.
- Require human approval for external messages, publishing, transactions, deployments, and destructive actions.
- Rotate and revoke an exposed credential immediately; deleting it from the latest revision is not sufficient because Git history may retain it.
No automated scan proves the absence of secrets or vulnerabilities. Contributors must combine repository checks, diff review, provenance review, and appropriate domain testing.
assets/banner.png
BIN
View File
Binary file not shown.
Side by Side Swipe Overlay

Before

Width:  |  Height:  |  Size: 106 KiB

After

Width:  |  Height:  |  Size: 836 KiB

assets/demo-install.cast
+14
View File
@@ -0,0 +1,14 @@
{"version":2,"width":86,"height":24,"timestamp":1784635200,"env":{"SHELL":"/bin/zsh","TERM":"xterm-256color"}}
[0.2,"o","\u001b[1;36mAGI Super Team — safe installer\u001b[0m\r\n\r\n"]
[0.8,"o","\u001b[1;37m$ ./install.sh solo-founder\u001b[0m\r\n"]
[1.4,"o","\u001b[36m[INFO]\u001b[0m Prerequisites passed\r\n\u001b[36m[INFO]\u001b[0m Deploying kit: solo-founder (3 agents)\r\n"]
[2.0,"o","\u001b[36m[PREVIEW]\u001b[0m CEO -> ./demo-workspace/workspace-ceo\r\n\u001b[36m[PREVIEW]\u001b[0m PE -> ./demo-workspace/workspace-pe\r\n\u001b[36m[PREVIEW]\u001b[0m CCO -> ./demo-workspace/workspace-cco\r\n"]
[3.2,"o","\u001b[1;32m[OK] PREVIEW complete. No files written.\u001b[0m\r\n\r\n"]
[4.0,"o","\u001b[1;37m$ ./install.sh --apply solo-founder\u001b[0m\r\n"]
[4.8,"o","\u001b[36m[INFO]\u001b[0m Required files passed global preflight\r\n\u001b[36m[INFO]\u001b[0m Staging selected agents\r\n"]
[6.0,"o","\u001b[1;32m[OK] Installed 3 agents without overwriting existing files\u001b[0m\r\n\r\n"]
[7.0,"o","\u001b[1;37m$ npm test && npm run validate -- --warnings-as-errors\u001b[0m\r\n"]
[8.1,"o","PASS workflows · structured files · manifest · symlinks · references · counts\r\n"]
[9.1,"o","\u001b[1;32m[OK] Repository and installer tests passed\u001b[0m\r\n"]
[10.2,"o","\u001b[1;32m[OK] 0 errors / 0 warnings\u001b[0m\r\n"]
[10.9,"o","\r\n\u001b[1;35mPreview. Apply. Verify.\u001b[0m\r\n"]
assets/demo-install.gif
BIN
View File
Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 74 KiB

assets/demo-install.txt
+20
View File
@@ -0,0 +1,20 @@
AGI Super Team installer demo transcript
Duration: approximately 11 seconds
Recorded behavior: preview, explicit apply, repository verification
$ ./install.sh solo-founder
[INFO] Prerequisites passed
[INFO] Deploying kit: solo-founder (3 agents)
[PREVIEW] CEO -> ./demo-workspace/workspace-ceo
[PREVIEW] PE -> ./demo-workspace/workspace-pe
[PREVIEW] CCO -> ./demo-workspace/workspace-cco
[OK] PREVIEW complete. No files written.
$ ./install.sh --apply solo-founder
[INFO] Required files passed global preflight
[INFO] Staging selected agents
[OK] Installed 3 agents without overwriting existing files
$ npm test && npm run validate -- --warnings-as-errors
[OK] Repository and installer tests passed
[OK] 0 errors / 0 warnings
assets/github-avatar.png
BIN
View File
Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 13 KiB

assets/logo-1024.png
BIN
View File
Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 349 KiB

assets/logo-mark.png
BIN
View File
Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 315 KiB

assets/logo.png
BIN
View File
Binary file not shown.
Side by Side Swipe Overlay

Before

Width:  |  Height:  |  Size: 59 KiB

After

Width:  |  Height:  |  Size: 349 KiB

assets/social-preview.png
BIN
View File
Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 612 KiB

setup.md
+56 -81
View File
@@ -1,116 +1,91 @@
# 🚀 Setup Guide — Agent Auto-Deploy
# Setup, verification, and recovery
> **如果你是 Claude Code / Codex / Cursor / Hermes 等 AI Agent，请按以下步骤操作。**
This guide covers the generic workspace installer. The curated Codex-native package is separate; follow [`.codex/INDEX.md`](./.codex/INDEX.md) for that distribution.
## 一键部署命令
## Prerequisites
告诉你的 Agent：
- Bash and standard Unix tools.
- Git when the installer must fetch a repository. A local `--source` avoids that fetch.
- Write permission for the selected `--destination`.
- A supported AI harness configured separately. The installer does not configure models, credentials, or provider accounts.
```
帮我安装这个 starter kit：https://github.com/aAAaqwq/AGI-Super-Team
OpenClaw CLI commands are legacy and optional. Codex, Claude Code, Cursor, Gemini, and Kimi have different extension models; repository metadata does not imply feature parity.
选择一个：
- solo-founder（一个人创业：CEO + PE + CCO）
- content-creator（自媒体：CCO + CDO + CMO）
- quant-trader（量化：CQO + CDO + CFO）
- full-team（全部12个Agent）
```
## Agent 执行步骤
当用户要求安装 AGI Super Team 时，按以下步骤执行：
> **推荐方式（harness 原生优先）**：
> - **Claude Code（推荐）**：直接运行 `/plugin install aAAaqwq/AGI-Super-Team`，无需本脚本。
> - **Codex / Cursor / Hermes 等**：使用下方 Step 1 的 `git clone` + `install.sh` 流程。
> - 下方依赖 OpenClaw CLI 的步骤（`openclaw config` / `openclaw gateway restart`）标注为
> *(legacy)*，仅当你的环境仍运行（已 discontinued 的）OpenClaw harness 时需要。
### Step 1: Clone & Deploy
## 1. Inspect a trusted checkout
```bash
# Clone 仓库
git clone --depth 1 https://github.com/aAAaqwq/AGI-Super-Team.git ~/.agi-super-team
# 运行安装脚本（选择对应 kit）
bash ~/.agi-super-team/install.sh solo-founder
# 可选：content-creator | quant-trader | full-team | ceo | pe | cco ...
git clone --depth 1 --branch main https://github.com/aAAaqwq/AGI-Super-Team.git
cd AGI-Super-Team
git rev-parse HEAD
```
### Step 2: Verify
Record the revision, inspect `install.sh`, and review the selected agent and skill directories before applying changes.
## 2. Preview
Preview is the default. Use an explicit destination so the proposed write locations are easy to audit.
```bash
# 检查 workspace 是否创建成功
ls ~/.openclaw/workspace-ceo/ # CEO
ls ~/.openclaw/workspace-pe/ # PE (如部署)
ls ~/.openclaw/workspace-cco/ # CCO (如部署)
./install.sh --source "$PWD" --destination /path/to/review-workspace solo-founder
```
每个 workspace 应包含：
- `SOUL.md` — 人格内核
- `AGENTS.md` — 工作手册
- `IDENTITY.md` — 身份档案
- `BOOTSTRAP.md` — 启动引导
- `skills/` — 精选技能
Other selectors are `content-creator`, `quant-trader`, `full-team`, or one agent ID such as `ceo`. A second positional agent ID filters a starter kit.
### Step 3: Configure API Keys
Preview may print an optional legacy OpenClaw warning. It should list planned agents and finish by asking you to re-run with `--apply`.
> *(legacy, if using OpenClaw harness)* — Claude Code / Codex / Cursor / Hermes 用户请在各 harness 的设置里直接配置 provider API key。
## 3. Apply after review
```bash
# 引导用户配置 API key (legacy, if using OpenClaw harness)
openclaw config
./install.sh --source "$PWD" --destination /path/to/review-workspace --apply solo-founder
```
需要至少一个 LLM provider 的 API key：
- Anthropic (Claude) — 推荐
- OpenAI (GPT-4)
- ZAI / Google / 其他
The installer creates `workspace-<agent>/` directories beneath the destination. It copies supported persona files and selected skill directories without replacing existing paths.
### Step 4: Restart & Test
Because existing files are preserved, repeated runs are not an upgrade mechanism for modified files. Review differences and merge intentionally.
> *(legacy, if using OpenClaw harness)* — 其他 harness（Claude Code / Codex / Cursor / Hermes）无需重启 gateway，直接在会话里使用即可。
## 4. Verify
```bash
# (legacy, if using OpenClaw harness)
openclaw gateway restart
find /path/to/review-workspace -maxdepth 2 -type f -name 'AGENTS.md' -print
find /path/to/review-workspace -maxdepth 3 -type f -name 'SKILL.md' -print
npm test
npm run validate
```
## 单 Agent 部署
Expected kit workspaces:
如果用户只想部署某一个 Agent：
| Kit | Expected directories |
|---|---|
| `solo-founder` | `workspace-ceo`, `workspace-pe`, `workspace-cco` |
| `content-creator` | `workspace-cco`, `workspace-cdo`, `workspace-cmo` |
| `quant-trader` | `workspace-cqo`, `workspace-cdo`, `workspace-cfo` |
Inspect the installed `SOUL.md`, `AGENTS.md`, `TOOLS.md`, and skills before trusting them. Some optional skills named by the installer may not exist and are reported as warnings.
## Codex-native package
The Codex distribution is under `plugins/agi-super-team-codex/` and registered by `.agents/plugins/marketplace.json`. It does not use `install.sh` or the starter-kit mapping.
Follow the commands and safe sync procedure in [the Codex package index](./.codex/INDEX.md). Agent sync previews first, backs up differing files, and requires explicit application.
## Updates
Fetch the `main` branch, inspect the diff, run repository checks, then preview the installer again:
```bash
# 部署单个 agent（支持别名和 ID）
bash ~/.agi-super-team/install.sh ceo # CEO
bash ~/.agi-super-team/install.sh pe # PE (开发)
bash ~/.agi-super-team/install.sh cco # CCO (内容)
bash ~/.agi-super-team/install.sh cto # CTO (架构)
bash ~/.agi-super-team/install.sh cqo # CQO (量化)
# ... 等等
git fetch origin main
git diff --stat HEAD..origin/main
git diff HEAD..origin/main -- install.sh agents skills starter-kits
```
## 自定义
Do not blindly pull and apply when local workspace files contain customizations. The installer intentionally preserves those files.
部署后用户可以自定义每个 Agent：
## Recovery
1. **改人格**：编辑 `~/.openclaw/workspace-{agent}/SOUL.md`
2. **加技能**：从 727 个 skills 中选择，复制到 `~/.openclaw/workspace-{agent}/skills/`
3. **改记忆**：编辑 `MEMORY.md` 注入领域知识
4. **加工具**：在 `skills/` 目录添加新 SKILL.md
If preview is wrong, stop and correct `--source`, `--destination`, or the kit selector; preview writes nothing.
### 推荐技能（按场景）
If an apply is interrupted, inspect the destination. Existing files were preserved, but newly created files may be partial. Move the newly created `workspace-<agent>` directory aside, confirm its exact path, then re-run preview and apply.
**独立开发者**：`thinking-elon-musk`, `thinking-linus-torvalds`, `api-design`, `docker-containerization`
**自媒体**：`khazix-writer`, `xhs-publisher`, `douyin-publisher`, `seo-writing`
**量化**：`backtesting-system`, `risk-management`, `data-pipeline`
**研究**：`deep-research`, `web-search`, `scientific-method`
If apply reports a missing required skill, do not substitute an unrelated skill. Confirm the checkout is complete and run `npm run validate` before retrying.
## 故障排查
| 问题 | 解决 |
|------|------|
| `openclaw: command not found` | *(legacy)* OpenClaw 已 discontinued；改用 Claude Code / Codex / Cursor / Hermes。如确需 OpenClaw：`npm install -g openclaw` |
| `Agent source not found` | 检查 clone 是否完整，重新 `git clone` |
| workspace 为空 | 检查 `~/.agi-super-team/agents/` 目录是否存在 |
| API key 无效 | *(legacy)* `openclaw config` 重新配置；或在对应 harness 设置里配置 provider API key |
Never paste credentials into repository files or command output. Configure providers through the chosen harness and keep secrets outside the project checkout.
starter-kits/content-creator/README.md
+15 -19
View File
@@ -1,26 +1,22 @@
# 🎨 Content Creator Starter Kit
# Content Creator starter kit
> 自媒体人的 AI 团队。内容创作 + 数据采集 + 营销增长，3 个 Agent 覆盖内容全链路。
This kit selects CCO for drafting, CDO for research and measurement, and CMO for positioning. It does not install the separate Codex-native package or publish to external platforms.
## 包含的 Agents
| Agent | 角色 | 精神导师 | 能做什么 |
|-------|------|----------|---------|
| ✍️ CCO | 内容创作 + 病毒传播 | MrBeast, Jony Ive | 小红书/抖音/公众号内容创作与发布 |
| 📊 CDO | 数据采集 + 分析 | Nate Silver | 竞品数据抓取、趋势分析、数据管线 |
| 📣 CMO | 营销 + SEO + 增长 | David Ogilvy | SEO优化、营销策略、竞品分析 |
## 一键部署
## Safe install
```bash
curl -sSL https://raw.githubusercontent.com/aAAaqwq/AGI-Super-Team/main/install.sh | bash -s -- content-creator
./install.sh --source "$PWD" --destination /path/to/review-workspace content-creator
./install.sh --source "$PWD" --destination /path/to/review-workspace --apply content-creator
```
## 你能让他们做什么？
Run the first command as a preview. Run the second only after reviewing the proposed `workspace-cco`, `workspace-cdo`, and `workspace-cmo` paths. Existing files and skill directories are preserved.
```
"抓取这个竞品账号最近30天的数据" → CDO
"把这篇内容同时发到小红书和抖音" → CCO
"分析这个关键词的SEO机会" → CMO
"帮我制定下周的内容日历" → CCO + CMO 协作
```
## Evaluation prompts
- CDO: “Summarize the supplied, licensed dataset and list missing evidence. Do not scrape private or restricted sources.”
- CCO: “Draft a post from the approved evidence pack. Mark unsupported claims and do not publish.”
- CMO: “Design a one-week ethical experiment with a hypothesis, audience, metric, stop condition, and manual review gate.”
Respect platform rules, consent, attribution, and community norms. All posts and replies must be reviewed and sent manually by an authorized human.
See [the setup guide](../../setup.md) and [growth playbooks](../../growth/README.md).
starter-kits/quant-trader/README.md
+17 -12
View File
@@ -1,17 +1,22 @@
# 📈 Quant Trader Starter Kit
# Quant Trader starter kit
> 量化交易员的 AI 团队。策略研发 + 数据工程 + 风控，3 个 Agent 覆盖量化全流程。
This kit selects CQO for quantitative research, CDO for data work, and CFO for risk and financial review. It is for research and paper-trading evaluation, not live execution or financial advice.
## 包含的 Agents
| Agent | 角色 | 精神导师 | 能做什么 |
|-------|------|----------|---------|
| 📈 CQO | 量化策略 + 交易 | Jim Simons | 策略开发、回测、风险管理、组合优化 |
| 📊 CDO | 数据采集 + ETL | Nate Silver | 市场数据抓取、清洗、特征工程 |
| 💰 CFO | 财务 + 风控 | Warren Buffett | 资金管理、P&L追踪、成本优化 |
## 一键部署
## Safe install
```bash
curl -sSL https://raw.githubusercontent.com/aAAaqwq/AGI-Super-Team/main/install.sh | bash -s -- quant-trader
./install.sh --source "$PWD" --destination /path/to/review-workspace quant-trader
./install.sh --source "$PWD" --destination /path/to/review-workspace --apply quant-trader
```
Run the first command as a preview. Apply only after reviewing the proposed `workspace-cqo`, `workspace-cdo`, and `workspace-cfo` paths. Existing files and skill directories are preserved.
## Evaluation prompts
- CDO: “Profile this licensed historical dataset for gaps, survivorship bias, leakage, and timestamp errors.”
- CQO: “Design a reproducible backtest with costs, slippage, walk-forward evaluation, and failure criteria.”
- CFO: “Review drawdown, concentration, liquidity, and operational risk. Do not recommend or place a trade.”
No bundled strategy is represented as profitable, live-validated, or production-ready. Use isolated test data, independent review, and human-controlled paper-trading before considering any further use.
See [the setup guide](../../setup.md) for verification and recovery.
starter-kits/solo-founder/README.md
+14 -84
View File
@@ -1,99 +1,29 @@
# 🚀 Solo Founder Starter Kit
# Solo Founder starter kit
> 一个人就是一支军队。CEO + 首席工程师 + 首席内容官，3 个 Agent 覆盖创业核心需求。
This kit selects three generic-workspace agents: CEO for planning, PE for engineering, and CCO for content drafts. It does not install the separate Codex-native package.
## 包含的 Agents
## Safe install
| Agent | 角色 | 精神导师 | 能做什么 |
|-------|------|----------|---------|
| 👑 CEO | 战略决策 + 团队协调 | Elon Musk | 第一性原理思考、任务调度、质量把控 |
| 💻 PE | 全栈开发 + DevOps | Linus Torvalds, DHH | 写代码、架构、Docker、测试、部署 |
| ✍️ CCO | 内容创作 + 病毒传播 | MrBeast, Jony Ive | 小红书/抖音/公众号发布、内容策略 |
## 一键部署
From a reviewed checkout, preview first:
```bash
# 方式1：curl（推荐）
curl -sSL https://raw.githubusercontent.com/aAAaqwq/AGI-Super-Team/main/install.sh | bash -s -- solo-founder
# 方式2：clone 后执行
git clone https://github.com/aAAaqwq/AGI-Super-Team.git
cd AGI-Super-Team
./install.sh solo-founder
./install.sh --source "$PWD" --destination /path/to/review-workspace solo-founder
```
## 部署后配置
Apply only after checking the planned paths and agents:
```bash
# 1. 设置 API Key（选择你的provider）
openclaw config
# 填入 API key，比如 Anthropic / OpenAI / ZAI
# 2. 重启 gateway
openclaw gateway restart
# 3. 开始使用
# CEO: 直接在配置的聊天通道跟它说话
# PE: 切换到 pe agent 或在群聊中 @它
# CCO: 切换到 cco agent
./install.sh --source "$PWD" --destination /path/to/review-workspace --apply solo-founder
```
## 你能让他们做什么？
The expected directories are `workspace-ceo`, `workspace-pe`, and `workspace-cco`. Existing files and skill directories are preserved.
### CEO (战略 + 协调)
```
"帮我分析一下这个竞品"
"制定下个月的产品路线图"
"review 一下这个技术方案的可行性"
```
## Evaluation prompts
### PE (开发 + 工程)
```
"帮我写一个 REST API，用 FastAPI"
"给这个项目写 Dockerfile 和 docker-compose"
"review 这个 PR 的代码质量"
"写一个 Playwright E2E 测试"
```
- CEO: “Draft a product decision memo with assumptions, alternatives, evidence gaps, and a human approval gate.”
- PE: “Propose a test-first implementation plan. Do not deploy or change production systems.”
- CCO: “Draft three launch-post variants with factual claim placeholders and a manual publishing checklist.”
### CCO (内容 + 传播)
```
"把这篇文章发到小红书"
"帮我写一篇关于 AI Agent 的公众号文章"
"分析一下这个账号的内容策略"
```
Review every output. Publishing, deployments, account access, and other external actions require explicit human authorization.
## 架构
```
你 (创始人)
└── 👑 CEO (ceo) — 接收你的指令，协调 PE 和 CCO
├── 💻 PE — 代码、技术、部署
└── ✍️ CCO — 内容、发布、传播
```
## 精选 Skills（已预装）
- **CEO**: team-coordinator, context-manager, healthcheck, daily-rhythm, web-search, project-planner
- **PE**: react-expert, tdd-workflow, systematic-debugging, code-review-quality, github, gh-issues, docker-containerization, deployment-automation, kubernetes-specialist, ghost-scan-code, cli-developer
- **CCO**: xhs-publisher, douyin-publisher, gzh-publisher, content-pipeline, seo-writing
## 扩展
想加更多 Agent？
```bash
# 加 CTO（架构设计）
./install.sh cto
# 加 CFO（财务）
./install.sh cfo
# 或者直接部署全部 12 个
./install.sh full-team
```
## 需要
- [OpenClaw](https://github.com/openclaw/openclaw) 已安装
- 至少一个 LLM API key（推荐 Claude / GPT-4）
- Telegram Bot（可选，用于聊天界面）
See [the setup guide](../../setup.md) for verification, updates, and recovery.