mesh  by decocms

Full-stack AI app development with context management

Created 10 months ago
304 stars

Top 88.2% on SourcePulse

GitHubView on GitHub
1 Expert Loves This Project
Project Summary

DecoCMS provides a unified, TypeScript-first platform for defining, composing, and deploying secure AI applications and workflows. It targets AI teams struggling with fragmented development stacks, offering a centralized "MCP Mesh" for context management, governance, and observability, enabling the creation of production-ready, full-stack AI applications deployable anywhere.

How It Works

DecoCMS centers around the MCP Mesh, a distributed runtime that aggregates AI model connections, tools, data access, and policies. It allows users to declare and compose Model Context Protocol (MCP) servers, exposing governed "Virtual MCPs" (AI Apps) to clients. The AI App Framework builds full-stack applications using React 19 and Tailwind v4, sharing types across agents, workflows, and UIs within a single TypeScript repository. This MCP-native, full-stack approach simplifies development and deployment, offering built-in governance and observability.

Quick Start & Requirements

  • Primary install/run: npm create deco, cd my-app, npm run dev (local development).
  • Prerequisites: Node.js/npm. Deployment targets include Cloudflare Workers, AWS, GCP, or local Bun/Deno runtimes. No specific GPU, CUDA, or Python version requirements are listed for core functionality.
  • Links: 📘 Docs, 💬 Discord, 🌐 decocms.com.

Highlighted Details

  • MCP-native: Connects AI models to data/tools via governed context.
  • Full-stack TypeScript: Unified runtime for backend (agents, workflows) and frontend (UIs) with typed RPC.
  • Deploy Anywhere: Edge-native via Cloudflare Workers or self-hosted on various runtimes.
  • Governance & FinOps: Built-in RBAC, audit trails, spend caps, and policy enforcement.
  • Unified Observability: End-to-end tracing (UI → agent → model) via OpenTelemetry, with cost analytics.
  • Deco MCP Store: Marketplace for reusing full-stack modules (agents, workflows, UIs).

Maintenance & Community

The project is driven by the "deco community" and welcomes contributions from "vibecoders" and "agentic engineers." Community interaction is facilitated via Discord. Upcoming milestones include a multi-tenant admin dashboard, MCP store enhancements, and an edge debugger.

Licensing & Compatibility

DecoCMS uses a Sustainable Use License (SUL). It is free to self-host and use for delivering client projects, provided each deployment is for the client's internal use (SUL §3.3). Commercial, self-hosted Enterprise licenses are required for mission-critical or revenue-generating applications, offering production support and scale tuning for AWS/GCP.

Limitations & Caveats

The SUL imposes specific restrictions on commercial use cases beyond internal client projects, requiring an enterprise license for revenue-generating or mission-critical deployments. The comparison tables suggest it is optimized for agentic systems and governed AI apps, rather than standard CRUD applications.

Health Check
Last Commit

1 day ago

Responsiveness

Inactive

Pull Requests (30d)
193
Issues (30d)
218
Star History
54 stars in the last 30 days

Explore Similar Projects

Starred by Elie Bursztein Elie Bursztein(Cybersecurity Lead at Google DeepMind), Chip Huyen Chip Huyen(Author of "AI Engineering", "Designing Machine Learning Systems"), and
7 more.

SuperAGI by TransformerOptimus

0.2%
17k
Open-source framework for autonomous AI agent development
Created 2 years ago
Updated 11 months ago
Feedback? Help us improve.