# ChengOS Official Launch Announcement: The Rust-Powered AI Workflow & Agent Platform is Live

# 🚀 ChengOS is officially live!

Today we're excited to announce the official release of **ChengOS** — a visual AI data processing and agent workflow platform built from scratch in Rust. If you've struggled with performance bottlenecks in Dify or n8n under high concurrency, or worried about handing your API keys to an LLM, ChengOS was built for exactly these problems.

## Why we built ChengOS

Over the past two years, AI workflow platforms have multiplied, but we kept running into three pain points:

1. **Performance ceiling**: most platforms are Node.js / Python — memory usage and latency suffer on complex DAGs and high concurrency
2. **Credential security**: stuffing API keys into prompts or environment variables means a single prompt injection can leak every secret
3. **Fragmented MCP ecosystem**: MCP (Model Context Protocol) is becoming the de facto standard for tool calling, yet most platforms treat it as a bolt-on, not a first-class citizen

ChengOS's answer: **Rust core + credential vault + native MCP**.

## Architecture at a glance

ChengOS uses a client-server architecture (REST API + WebSocket) with four modules:

| Module | Stack | Responsibility |
|---|---|---|
| **chengflow** | Rust | Core DAG workflow engine, REST/WS API, credential vault, multi-tenant RBAC |
| **chengflow-ui** | React 18 + TypeScript | Visual console: React Flow editor, MCP Hub, skill orchestration |
| **chengflow-app** | React 18 + TypeScript | Chat gateway for Telegram / WhatsApp / Slack / WeCom / DingTalk |
| **cheng CLI** | Rust | Terminal client with a Claude Code-style interactive experience |

## Six core capabilities

### 1. Native MCP support

Built-in MCP client and server, supporting Stdio, SSE, and HTTP transports. Let agents in your workflows call any external MCP server — or expose ChengOS itself as an MCP service to other AI tools.

### 2. Four ways to import skills

- Natural language description → auto-generated skill
- URL import: paste a link and it gets parsed
- Git repository parsing: extract capabilities straight from a repo
- SKILL.md standard import: compatible with Anthropic's skill spec

### 3. Credential vault

Credentials are stored encrypted and **the LLM never touches raw keys** — the engine injects them on tool invocation, structurally preventing key leaks via prompt injection.

### 4. Local document workspace

Built-in Word / Excel visual canvases — no need for Feishu or Google Sheets, your data never leaves your machine.

### 5. CLI experience

The `cheng` CLI offers full terminal interaction with remote URL connection — drive your workflows from a server or an SSH session as naturally as Claude Code.

### 6. Multiple chat entry points

Web embed, App H5 / PWA, and CLI — plus chengflow-app to plug agents into Telegram, WeCom, DingTalk and other IMs.

## 90+ built-in nodes, 9 model providers

Out of the box: code sandbox, file management, web scraping, headless browser, image/audio/video generation, ASR, OCR, RAG knowledge base, scheduled triggers… 90+ nodes cover most automation scenarios.

On the model side: OpenAI, Claude, Gemini, DeepSeek, Ollama and 9 providers in total — local and cloud models can be mixed.

## One-line deployment

```bash
curl -fsSL https://chengos.dev/chengos.sh | bash
```

Docker or binary installation; the script bundles PostgreSQL, Valkey, and the Qdrant vector DB automatically. Full instructions in the [Installation Guide](/docs/install-chengos).

## License and editions

ChengOS is currently **free to use but not open source**. It may be used for personal and commercial office automation, but the source code is not publicly available.

Planned editions:

- **Community Edition** — free, with the current core automation capabilities
- **Enterprise Edition** — paid, with team collaboration, permission management, analytics, and additional enterprise features

For business inquiries or enterprise consultation, contact hello@chengrouter.com.

## Join the community

ChengHub is the official ChengOS community — you're already here! You can:

- 💡 Post **requirements** and vote; the team tracks status transitions
- 📝 Share **experiences** — tutorials and lessons learned are welcome
- 🔀 Share **workflow** templates others can import in one click
- 💬 Join **discussions**, ask questions, show off your projects

QQ group: **1090876074** — come chat with us.

Bonus: ChengHub natively supports MCP access — browse docs, search the community, and post from Claude Code, Cursor and other MCP clients. See the [Welcome post](/posts/1f8fd1ea-b590-4102-8a80-99d0778af038) for setup.

## Roadmap preview

- More built-in nodes (video processing, data analysis)
- Workflow marketplace: one-click community templates

---

**Get started now:** after installing, reply to this post and tell us what you plan to build with ChengOS. We read every reply.