MCP Catalogue

183 open source services. One MCP connection.

Every major AI cloud now speaks MCP. What separates OSC is the catalogue — 183 vetted open source services with source code, zero runtime lock-in. Deploy through natural language with any MCP-compatible AI tool.

Eyevinn Open Source Cloud (OSC) is a managed platform for 183+ open source services, accessible via MCP server at mcp.osaas.io/mcp. Enable vibe deployment from any AI coding tool. MCP is the protocol layer for agentic engineering — giving AI agents the infrastructure they need to build and deploy autonomously.

Try in Browser

Why managed MCP

Why managed MCP deployment matters

Self-host means operating containers, secrets, ingress, and TLS yourself. Closed-source platforms lock you to one cloud runtime. OSC is the managed open-source option, every catalog service runs the unmodified upstream container, so workloads stay portable.

Self-hosted MCP

You operate everything: container orchestration, secrets, ingress, TLS, scaling, upgrades. Full control, full ops burden.

Closed-source managed MCP

No ops, but the runtime is proprietary. Migration off the platform means re-platforming. Your services run on a vendor stack, not open source.

Open Source Cloud (managed open source MCP)

Managed for you. Every service runs the unmodified upstream open-source container. Move workloads to any cloud or on-premises any time. No proprietary runtime, no licensing cost.

Quick Start

Start Building with AI on OSC

One command connects your AI agent to 183+ managed open source services. Your agent provisions infrastructure as you code.

MCP endpoint:https://mcp.osaas.io/mcp

Other tools: Windsurf, Amazon Q CLI, JetBrains, Cline, Warp — see full setup guides

Tenant-aware

Live tenant inventory in your AI agent

Once connected, your AI agent immediately sees your running services and apps, not just a generic catalog. list-my-services, list-my-apps, and per-instance endpoints with live credentials return scoped to your tenant. Your agent can manage what you already have, not just create new things.

Example agent prompt:

> Restart my staging Postgres, take a backup of the production one, and tell me what storage buckets I have.

Agent uses: list-my-services, restart-service-instance, create-backup, list-objects-on-bucket, all scoped to your tenant, no IDs to look up.

Production-Ready, Not Sandboxed

Other tools let you prototype in a sandbox. OSC MCP deploys real infrastructure from the start.

Sandbox-then-Export

OSC MCP

Workflow

Plan → Sandbox → Export IaC → Configure CI/CD → Deploy

Describe → Deployed

Time to production

Hours / days

Minutes

Lock-in

Requires specific IaC tooling

Zero lock-in (open source)

Transparency

Black-box sandbox

Full source code access

See the proof

Streaming Tech TV+ — 14 services in 36 hours Specter CRM — full CRM in a weekend

MCP Tools

Access the complete OSC infrastructure through natural language

Model Context Protocol (MCP) gives AI agents structured access to 183+ open source services. Zero vendor lock-in — every service is portable open source.

Infrastructure

Databases, storage, caching, message queues

Applications

Deploy and manage complete applications

Intelligence

AI-powered architecture planning and optimization

Zero Lock-In

Every deployed service is unmodified open source. Move to any cloud at any time.

Ask OSC Architect

Get expert guidance on building complex architectures

Host Your MCP Server on OSC

Your MCP server, production-ready in minutes. HTTPS, SSE, auto-scaling — no DevOps required.

Deploy from GitHub

Any Node.js, Python, or Wasm MCP server. Just point us to your repo.

HTTPS Endpoint Instantly

Get a public URL like https://{app}.auto.prod.osaas.io/mcp — share it with anyone.

Auto-Scaling, Zero Lock-In

Your code stays yours. Deploy anywhere, no vendor lock-in.

Deploy Your MCP Server |Read the Tutorial

Connect Your AI Agent in 2 Minutes

Connect to Open Source Cloud today and unlock the full potential of open source media tools with your favorite AI assistant.

Try in Browser

Featured Tutorial

15 min

Build Your First App

Go from zero to a deployed full-stack application using Claude Code and Open Source Cloud.

Start the Tutorial

Free tier: 100 tokens, no credit card. Paid plans from EUR 15/mo. View pricing →

Try: "Create a PostgreSQL database" with your AI assistant

Frequently Asked Questions

Run a single command in your terminal: claude mcp add --transport http osc https://mcp.osaas.io/mcp. For Cursor, VS Code, or other tools, add the MCP server URL https://mcp.osaas.io/mcp in your MCP configuration with a Bearer token from your free OSC account at app.osaas.io. Once connected, your AI agent can discover and deploy any of 183+ managed open source services through natural language.

Connect AI Agents to Open Source Cloud via MCP

Open Source as a Service (OSaaS) is a cloud model where every service is an unmodified open source project. With the OSC MCP server, your AI agent can deploy any of 183+ OSaaS services — databases, video pipelines, media processing, AI tools — through natural language, with zero vendor lock-in.

MCP Server Endpoint

https://mcp.osaas.io/mcp

Use this endpoint with any MCP-compatible AI client. Authentication requires a Bearer token (personal access token from app.osaas.io).

Compatible AI Clients

OSC works with every major MCP-compatible AI client. The universal endpoint above is the same for all. Pick your agent:

Claude Desktop App — Open Settings → Connectors → “Add custom connector”, name it “OSC”, and paste the MCP URL https://mcp.osaas.io/mcp. Authorize when prompted.
Claude Code (CLI) — Run claude mcp add --transport http osc https://mcp.osaas.io/mcp, then /mcp inside Claude Code triggers the OAuth flow.
Codex (OpenAI CLI) — Run codex mcp add osc --url https://mcp.osaas.io/mcp, then codex mcp login osc to authenticate via OAuth.
Cursor — One-click install deeplink on the wizard, or add to .cursor/mcp.json with "url": "https://mcp.osaas.io/mcp".
VS Code (GitHub Copilot) — Command Palette → MCP → “Add server” → Streamable HTTP → paste the URL.
ChatGPT — Settings → Connectors → toggle Developer Mode (requires Pro/Team/Enterprise/Edu), then in chat click + → Developer Mode → Add sources → paste the MCP URL.
Windsurf, Amazon Q CLI, JetBrains AI Assistant, Cline, Warp — Edit each tool's mcp_config.json with the universal URL above. Full per-tool snippets in the wizard.
GitHub Actions — Pass MCP config to anthropics/claude-code-action via --mcp-config.

MCP Configuration (JSON)

{
  "mcpServers": {
    "osc": {
      "type": "http",
      "url": "https://mcp.osaas.io/mcp",
      "headers": {
        "Authorization": "Bearer <personal-access-token>"
      }
    }
  }
}

Available Capabilities

List and search available services
Create and manage service instances
Create databases (PostgreSQL, MySQL, CouchDB, Valkey)
Deploy custom applications from GitHub repositories
Create storage buckets
Manage secrets and configuration
Deploy solution templates
Ask the OSC Architect for help building solutions

Why managed MCP deployment matters

Self-hosting MCP servers means operating containers, secrets, ingress, and TLS yourself. Closed-source platforms lock you to one cloud's runtime. OSC is the managed open-source option: your code runs in unmodified open-source containers, so you can move workloads to any cloud or on-premises without re-platforming. 183+ services. One MCP endpoint.

Live tenant inventory in your AI agent

When you connect OSC MCP, your agent immediately sees your running services and apps — not a generic resource catalog. list-my-services, list-my-apps, and per-instance endpoints + credentials are returned scoped to your tenant. Your agent can manage what you already have, not just create new things.

Get Started

1. Create an account at app.osaas.io (free tier: 100 tokens, no credit card)
2. Add the MCP server to your AI client using the install snippets above — OAuth handles the auth on Claude Desktop, Claude Code, Codex, and Cursor; other clients use a personal access token from app.osaas.io → Settings → API
3. Ask your agent to deploy or manage any of the 183+ services.

Read the full documentation

183 open source services. One MCP connection.

Why managed MCP deployment matters

Self-hosted MCP

Closed-source managed MCP

Open Source Cloud (managed open source MCP)

Start Building with AI on OSC

Live tenant inventory in your AI agent

Production-Ready, Not Sandboxed

MCP Tools

Infrastructure

Applications

Intelligence

Zero Lock-In

Ask OSC Architect

Host Your MCP Server on OSC

Deploy from GitHub

HTTPS Endpoint Instantly

Auto-Scaling, Zero Lock-In

Connect Your AI Agent in 2 Minutes

Build Your First App

Frequently Asked Questions

How do I connect my AI coding agent to Open Source Cloud?

What can I build on Open Source Cloud with an AI agent?

Is it free to use MCP with Open Source Cloud?

How do I deploy a full-stack app built with Claude Code?

What is managed MCP deployment?

Can I move my MCP servers and apps to another cloud later?

Do I need a hyperscaler cloud account to use OSC MCP?

What is the difference between hosted MCP and self-hosted MCP?

Will my AI agent see my existing services, or only new ones I create?

What is Model Context Protocol (MCP)?

Is the OSC MCP server free to use?

Which AI agents work with OSC?

What can I do through the MCP interface?

How do I get started with MCP?