TreeMap
무료 오픈소스 디스크 사용량 시각화 도구로, 사각형 트리맵·중복 파일 탐색·디스크 가득 예측·정리 시 파일을 휴지통으로만 이동한다
Praxis는 최첨단 인텔리전스를 활용해 아이디어를 실현 가능한 솔루션으로 전환하는 자율 문제 해결·프로젝트 관리 머신입니다
Harnessing frontier intelligence to turn ideas into practical reality. More than a personal assistant, Praxis is an autonomous problem-solving and project-management machine that evolves itself and th
생산성JavaScript · TypeScriptGitHub @VibeShiftAI온라인운영자 미확인
써 보셨어요? 한 번만 눌러 주세요. 로그인은 필요 없어요.
개인 계정의 공개 저장소 소유자는 GitHub로 로그인해 운영자를 확인하고 소개를 직접 관리할 수 있습니다. 관리자 승인은 필요하지 않습니다.
로그인한 뒤 확인 파일을 만들어 이 저장소의 기본 브랜치에 30분 안에 커밋해 주세요. 다른 브랜치나 포크의 파일은 인정하지 않습니다. 파일을 올린 뒤 아래에서 확인하면 관리자 승인 없이 관리할 수 있습니다. 자격을 유지하려면 파일을 남겨 두세요.
로그인이 필요하면 로그인 화면으로 이동합니다. 로그인 후 이곳으로 돌아와 원하는 확인 버튼을 다시 눌러 주세요.
프로젝트 폴더에서 /nomorevibe verify를 실행하면 소유권을 확인하고 소개를 직접 관리할 수 있습니다.
1분 만에 확인하기Personal Agentic Workspace — A local, open-source command center orchestrating 26+ AI agents to turn ideas into code.
저장소 README에서 가져옴
README 전체를 불러오는 중입니다. 아래는 저장된 발췌입니다.
Personal Agentic Workspace — Turns your ideas into real, manageable projects on your computer. 26+ specialized AI agents handle everything from research to code to security — you manage the vision, not the syntax. 🌐 Website | 🎥 Watch the Announcement Video | ☕ Support on Patreon ⚠️ Disclaimer: This project was entirely vibecoded by a single person. It is a personal experiment in AI-assisted development and is not intended for production use. Expect rough edges, unconventional patterns, and the occasional "it works on my machine" moment. Use at your own risk — and have fun with it.README 전문 보기 ↗
Personal Agentic Workspace — Turns your ideas into real, manageable projects on your computer. 26+ specialized AI agents handle everything from research to code to security — you manage the vision, not the syntax.
🌐 Website | 🎥 Watch the Announcement Video | ☕ Support on Patreon
⚠️ Disclaimer: This project was entirely vibecoded by a single person. It is a personal experiment in AI-assisted development and is not intended for production use. Expect rough edges, unconventional patterns, and the occasional "it works on my machine" moment. Use at your own risk — and have fun with it.
The Nexus bridges your local filesystem with an intelligent web dashboard, turning your laptop into a connected development fortress. Every decision passes through a human-in-the-loop review gate, so you manage the project, not the code.
/Volumes/Projects/Backup/TheNexus-dead-weight-2026-09-03)services/praxis-mind-mcp/), the Model Context Protocol surface AI agents use to read and govern the boardgraph TD
User["👤 You"] -->|natural language| Terminal
subgraph Frontend ["🖥️ Dashboard — Next.js :3000"]
Terminal["💬 AI Terminal"]
Tasks["📋 Task Manager"]
Agents["🤖 Agent Manager"]
Monitor["📊 System Monitor"]
Codex["📖 The Codex"]
end
subgraph Backend ["⚡ Node.js API — Express :4000"]
REST["REST API\n105 endpoints"]
Discovery["🔍 SOTA Model\nDiscovery"]
Critic["🔬 Code Critic"]
Supervisor["👔 Supervisors"]
end
subgraph Persistence ["💾 Data"]
DB[("SQLite\nnexus.db")]
FS[("Local\nFilesystem")]
end
subgraph Providers ["☁️ AI Providers"]
Gemini["Google Gemini"]
Claude["Anthropic Claude"]
GPT["OpenAI GPT"]
Grok["xAI Grok"]
end
Terminal --> REST
Tasks --> REST
REST --> Supervisor
Discovery --> Providers
Supervisor --> Providers
REST --> DB
REST --> FS
Critic -.->|review before write| FS
style User fill:#0f172a,stroke:#06b6d4,stroke-width:2px,color:#06b6d4
style Terminal fill:#0f172a,stroke:#06b6d4,stroke-width:2px,color:#67e8f9
style Discovery fill:#0f172a,stroke:#10b981,stroke-width:2px,color:#6ee7b7
style Critic fill:#0f172a,stroke:#8b5cf6,stroke-width:2px,color:#c4b5fd
style DB fill:#1e293b,stroke:#64748b,stroke-width:2px,color:#94a3b8
From your idea to deployed code — the 8-node Project Plan Generator orchestrates every step. Infrastructure includes the Blackboard (shared memory) and Glass Box Broadcasting (WebSocket artifacts for real-time visibility). See it live →
TheNexus/ # Flat monorepo
├── server/ # Node.js Express backend
│ ├── server.js # Main API server
│ ├── scanner.js # Project discovery engine
│ ├── agent/ # Multi-provider AI agent
│ ├── routes/ # Modular route handlers
│ ├── services/ # Business logic services
│ │ ├── model-discovery.js # SOTA model auto-detection
│ │ ├── critic.js # Code review service
│ │ └── ... # 10+ services
│ └── tools/ # Agent tool definitions
├── dashboard/ # Next.js 16 frontend
│ └── src/
│ ├── app/ # App Router pages
│ ├── components/ # UI components
│ └── lib/nexus.ts # Centralized API client
├── cortex/ # Python AI Brain (Pipeline orchestrator)
├── config/ # Centralized configuration
│ ├── model_registry.yaml # LLM model configs
│ └── prompts.yaml # System prompts
├── db/ # SQLite schema & migrations
├── .context/ # Project context docs (for AI agents)
└── Start The Nexus.example.bat # Windows startup template (double-click)
└── Start The Nexus.example.command # macOS startup template (double-click)
└── start-local.example.sh # Linux startup template
Archived on 2026-09-03 (no longer in the tree): _retired/nexus-builder/, sandbox/, tests/cortex/, antigravity-extension/ — moved to /Volumes/Projects/Backup/TheNexus-dead-weight-2026-09-03.
Start by right clicking on your main "Projects" folder and click "Open in Terminal", then copy and paste the following commands one by one. Your main Projects folder should not be nested inside Documents, Downloads, or any other system folders.
git clone https://github.com/VIbeShiftAI/TheNexus.git
cd TheNexus
install-windows.bat
chmod +x install-mac.sh
./install-mac.sh
The installer handles everything: Node.js dependencies, Dashboard build dependencies, Python virtual environment, and configuration file scaffolding.
After installation, configure your API keys. Note, I can't guarantee you can get full functionality without both a Google and Anthropic API key, as those were the 2 I started development with:
Option A — Dashboard Settings (recommended)
Once The Nexus is running, click the ⚙️ Settings gear icon in the top-right of the dashboard. The settings panel lets you edit your PROJECT_ROOT and all API keys — changes are saved to both .env files automatically.
Option B — Command Line
setup-env.bat # Windows
./setup-env.sh # macOS / Linux
Double-click Start The Nexus.bat in Explorer, or run from terminal:
"Start The Nexus.bat"
Double-click Start The Nexus.command in Finder, or run from terminal:
open "Start The Nexus.command"
./start-local.sh
Opens 2 terminal windows:
| Endpoint | URL |
|---|---|
| Dashboard | http://localhost:3000 |
| Node.js API | http://localhost:4000 |
Configure via the ⚙️ Settings page in the dashboard, or edit .env directly:
# Required
PROJECT_ROOT=/path/to/your/projects
# AI Providers (at least one)
GOOGLE_API_KEY=your-key
ANTHROPIC_API_KEY=your-key
OPENAI_API_KEY=your-key
XAI_API_KEY=your-key
# Frontend
NEXT_PUBLIC_API_URL=http://localhost:4000
NEXT_PUBLIC_CORTEX_URL=http://localhost:8100
On startup, the Model Discovery Service (server/services/model-discovery.js) queries the model listing APIs of all 4 providers in parallel, matches models against known families (Gemini Pro, Claude Opus, GPT, Grok), and selects the highest version per family — zero-config model upgrades.
The former Python workflow orchestration service was decommissioned on 2026-07-02 and moved to _retired/nexus-builder/, which was itself archived out of the repository on 2026-09-03 to /Volumes/Projects/Backup/TheNexus-dead-weight-2026-09-03. See docs/decommission/langgraph-2026-07-02.md for the decision record. It is preserved for recovery only and is not part of the live startup path.
graph LR
Idea["💡 Idea"] -->|"Gemini Deep Research"| Researching["🔬 Researching"]
Researching -->|"Background Polling\n4hr max"| Researched["📋 Researched"]
Researched -->|"👤 Approve"| Planning["📐 Planning"]
Researched -.->|"👤 Reject"| Idea
Planning -->|"Architect Agent"| Planned["✅ Planned"]
Planned -->|"👤 Approve"| Implementing["🔨 Implementing"]
Planned -.->|"👤 Reject"| Idea
Implementing -->|"Builder Agent\nwith Tools"| Testing["🧪 Testing"]
Testing -->|"👤 Approve"| Complete["🚀 Complete"]
Testing -.->|"👤 Reject"| Planned
Testing -.->|"👤 Cancel"| Cancelled["🗑️ Cancelled"]
style Idea fill:#334155,stroke:#94a3b8,color:#e2e8f0
style Complete fill:#14532d,stroke:#22c55e,color:#22c55e
style Cancelled fill:#450a0a,stroke:#ef4444,color:#ef4444
Each stage has human-in-the-loop gates. Research runs asynchronously with automatic resume on server restart.
graph TD
subgraph Dashboard ["Dashboard Level"]
Init["📋 Dashboard Initiative\nSecurity Sweep, Dependency Audit, etc."]
end
subgraph Projects ["Project Level"]
PA["Project A"]
PB["Project B"]
PC["Project C"]
end
subgraph Features ["Feature Level"]
FA1["Task 1\nResearch → Plan → Build"]
FA2["Task 2\nResearch → Plan → Build"]
FB1["Task 1\nResearch → Plan → Build"]
end
Init --> PA
Init --> PB
Init --> PC
PA --> FA1
PA --> FA2
PB --> FB1
style Init fill:#422006,stroke:#f59e0b,stroke-width:2px,color:#fbbf24
style PA fill:#1e1b4b,stroke:#8b5cf6,color:#c4b5fd
style PB fill:#1e1b4b,stroke:#8b5cf6,color:#c4b5fd
style PC fill:#1e1b4b,stroke:#8b5cf6,color:#c4b5fd
Current dashboard work flows through the Node.js API and task board. Historical project workflow diagrams remain in the docs for reference but should not be treated as live startup or routing guidance.
| Tool | Description |
|---|---|
read_file | Read file contents with offset/range support |
write_file | Create or overwrite file (with Critic review) |
patch_file | Replace specific text in file |
append_file | Append content to end of file |
apply_diff | Apply unified diff for targeted edits |
edit_lines | Edit specific lines by line number |
list_directory | List directory contents |
run_command | Execute shell command in project directory |
check_ports | List active listening ports |
kill_process | Terminate a process by PID or port |
checkpoint_memory | Save checkpoint context for long-running tasks |
The Codex (/codex) is the documentation hub with interactive visualizations:
| Section | Description |
|---|---|
| End-to-End Data Flow | SVG diagram of the full 8-node pipeline |
| Vibecoding Workflow | Interactive diagram of all 4 fleets with sub-agents |
| Initiative Hierarchy | Dashboard → Projects → Tasks cascade |
| Interface Overview | Annotated screenshots of Dashboard and Project views |
| Agent Registry | Live browsable registry of all agents and fleets |
| Component | Description |
|---|---|
ai-terminal.tsx | Multi-provider AI chat interface |
task-manager.tsx | Task management with status pipeline |
task-detail-modal.tsx | Full task workflow UI |
agent-manager.tsx | Configure AI agents |
resource-monitor.tsx | System monitor (CPU/memory/ports) |
activity-feed.tsx | Recent commits across projects |
dashboard-initiatives.tsx | Cross-project initiative management |
project-card.tsx | Project tile with git status |
project-context-manager.tsx | Project context document editor |
project-workflows.tsx | Project-level workflow management |
server/tools/ with Zod schemaserver/tools/index.jsAdd a pattern to MODEL_FAMILIES in server/services/model-discovery.js — the service auto-detects the latest version at next startup.
Detailed architectural docs live in:
.context/ — API reference, workflow maps, pipeline architecture, node reference, tech stackdocs/ — System architecture, data flow diagramsdocs/verification-protocol.md — How TheNexus work is briefed, verified, and reviewed (the protocol on file)/codex in the dashboardThe Nexus stands on the shoulders of outstanding open-source projects:
@xyflow/react) — Interactive node-based visualizations in The CodexThis project is licensed under the MIT License.