Introduction

What is deco CMS?

Deco CMS is an open-source Context Management System, the MCP Mesh for AI. It enables teams to build, deploy, and manage AI-native applications through unique collaboration between business users and developers.

Think Lovable + n8n + LangGraph, running on Cloudflare with a single deploy command. Full-stack AI, production-ready.

Value Proposition

deco CMS bridges the gap between AI prototypes and production systems:

• For AI Builders: Natural-language builder for business users to prototype and deploy AI apps without code
• For Context Engineers: Open-source TypeScript SDK for developers to extend, customize, and self-host
• Seamless Collaboration: Both users work together in the same platform with shared context

Architecture at a Glance

deco CMS is built on two foundational layers:

1. MCP Mesh (The Kernel)

The backbone that manages context, connections, and observability for everything in your system.

• Compose and secure MCPs across your organization
• Connect and proxy external MCPs with secure tokens
• Expose governed Virtual MCPs (“AI Apps”) to any MCP client
• Enforce auth/RBAC/audit/FinOps at every layer
• Get full observability across the entire stack

2. AI App Framework (Virtual MCPs in the Mesh)

Build AI-native web software that calls tools:

• Full-stack: From database to UI with React 19 + Tailwind v4
• Generative Admin: deco chat helps you write PRDs and implement
• Admin, workflows and custom views governed by the Mesh

Core Building Blocks

Tools

Functions that perform actions like API calls, database queries, or computations.

 createTool({
  id: "CEP_SEARCH",
  description: "Search Brazilian postal codes",
  inputSchema: z.object({ cep: z.string() }),
  outputSchema: z.object({ city: z.string(), state: z.string() }),
  execute: async ({ context }) => { /* ... */ },
}); 

Agents

AI assistants that understand natural language and call tools. Customizable with:

• Model selection (GPT-5, Claude, etc.)
• System prompts
• Tool access
• Resource context

Workflows

Deterministic tool sequences for automation. Unlike agents that decide, workflows follow predefined steps.

 createWorkflow({ id: "REGISTER_CUSTOMER" })
  .then(validateInput)
  .then(saveToDatabase) 
  .then(sendWelcomeEmail)
  .commit(); 

Views

Custom React interfaces for your tools and workflows. Built with React 19 + Tailwind v4.

Documents

Markdown content (PRDs, guides, notes) attachable to agents / any thread for context.

Resources

Everything in deco CMS is a resource: tools, agents, workflows, views, and documents. Resources are:

• Typed with Zod schemas
• Stored in deconfig (git-like versioning)
• Referenceable using @ mentions in threads and agent system prompts

Database

Built-in SQLite with Drizzle ORM for data persistence.

MCP Mesh

Context aggregation layer that unifies integrations, tools, and policies:

• External MCP proxy
• Secure credential storage
• RBAC and policy routing
• OpenTelemetry tracing
• Cost tracking

Who is deco CMS for?

AI Builders: Product managers, ops teams, business analysts building apps without code

Context Engineers: Developers extending with TypeScript, self-hosting, building custom integrations

Why deco CMS?

AI teams are stuck between low-code prototypes and production chaos. Backends in n8n or LangGraph. Frontends in Lovable or React. Separate deployments. Inconsistent auth. Spiraling costs.

deco CMS unifies everything:

• 🧠 MCP-native, Compose servers with built-in policy, auth, observability
• ⚙️ Full-stack TypeScript, Agents, workflows, UIs in one repo
• 🌍 Deploy anywhere, Cloudflare Workers, AWS, or local runtimes
• 🔐 Governance built-in, RBAC, audit trails, spend caps
• 🔭 Unified observability, Trace UI → agents → models
• 🧩 Open & modular, Install from deco store or build your own

Ready to build your first AI app? Choose your path:

• For AI Builders - Build without code
• For Context Engineers - Develop with TypeScript