Give your AI agent a backend it can deploy, and you can walk away with
Your AI coding agent can build the app. OSC gives it a backend to deploy through one MCP connection, built from unmodified open-source services you can export and run yourself, on any cloud, any time.
You told your coding agent to build the app. It wrote the code. Now it needs a database, object storage, and somewhere to actually run, and you do not want to spend the afternoon learning Kubernetes or wake up next month locked into whatever proprietary backend the platform quietly picked for you. That is the job: let the same agent provision real infrastructure, over a standard protocol, using software you could run yourself if you ever had to. Eyevinn Open Source Cloud is an MCP server that does exactly that. Point your agent at it and it can create a PostgreSQL database, an S3-compatible object store, a message queue, or a full backend-plus-hosting stack, by conversation. Every service it provisions is an unmodified upstream open-source container, not an OSC-proprietary fork, so the stack it hands you is one you actually own afterward, not one you are renting access to. Adding the server takes one line: claude mcp add --transport http osc https://mcp.osaas.io/mcp. Then run /mcp to authorise over OAuth 2.1 with PKCE. No API key to paste, no token file to manage. The catalog behind it holds 188+ published services across 8 categories, media being the largest.
Why This Is the Question to Ask in 2026
Your AI coding agent can write the app. The open question is where the backend runs afterward and who owns it. Write-capable cloud-management MCP servers are common now, so the fact that an agent can call a deploy tool stopped being the differentiator months ago. What matters is what the agent provisions for you: an open stack you can take with you, or a managed surface over one vendor's proprietary services that you can describe but not leave. OSC provisions the backend from unmodified open-source services and then lets you leave with all of it. That second half is the part a locked-in platform cannot truthfully offer, because there is nothing underneath its managed surface for you to take.
How to Add OSC to Your Agent
Four steps, and the first one is a single line. One, add the server: claude mcp add --transport http osc https://mcp.osaas.io/mcp. Two, authorise by running /mcp in Claude Code, which triggers OAuth 2.1 with PKCE against the OSC auth server. Three, describe the job rather than the parts. For example: I need a database and object storage for the app you just built, and I want to be able to run it myself later if I ever leave. The agent calls OSC MCP tools to create the services and returns the connection details. Four, wire your app to the returned connection strings, then deploy the app code itself from GitHub onto OSC, or run it anywhere you like. You own the services either way. OSC's MCP endpoint is a standard streamable-HTTP server. Claude Code, Claude Desktop, Codex, Cursor, VS Code with GitHub Copilot, and ChatGPT Developer Mode can all connect to it. Claude Desktop, Claude Code, Codex and Cursor authorise over OAuth; other clients use a personal access token from app.osaas.io.
What "You Own It" Actually Means
Every service OSC provisions is unmodified open source running in a standard container. There is no OSC-proprietary fork in the middle. The Exit Test is the runnable proof: provision a service, export the data with curl, run the identical container on your own machine with Docker, and diff the data. It is the same bytes. That is what no runtime lock-in means in practice, not as a marketing phrase but as a script you can run yourself, before you commit to anything.
Frequently Asked Questions
What's the actual job this solves?
Your AI coding agent wrote an app and now needs a real backend to run it against. OSC lets the same agent provision that backend, database, storage, queues, hosting, over MCP, and hands you infrastructure you can verify and leave with, instead of a managed surface you can only describe.
Can my AI agent provision a database through this MCP server?
Yes. The agent calls a tool over the OSC MCP endpoint to create the service, for example a PostgreSQL or PostgreSQL-with-pgvector instance, and gets back the connection details to wire into your app.
Do I own what the agent deploys?
Yes. Every provisioned service is unmodified open source in a standard container. There is no proprietary runtime layer between you and the software, and the Exit Test demonstrates that end to end.
Can I run the stack without OSC later?
Yes. Because the services are standard open-source containers, you can export your data and run the same images yourself. The Exit Test script walks through exactly that.
Does it work with agents other than Claude Code?
Yes. The OSC MCP endpoint works with Claude Desktop, Claude Code, Codex, Cursor, VS Code with GitHub Copilot, and ChatGPT Developer Mode, any MCP client that supports URL-based streamable-HTTP servers.
Is it free to connect?
Yes. Connecting your agent to OSC via MCP is free. You only pay for the services your agent actually deploys, on standard pay-as-you-go pricing, with a free tier that includes the full catalog.
Related Posts
Add an MCP Server to Your Stack in One Line
Connect your AI coding agent to real infrastructure. Add the Open Source Cloud MCP server in one line and let your agent provision and run open-source services.
Deploy What Your AI Just Built
Your AI coding assistant wrote the app. Here is how to deploy the backend on Open Source Cloud in one line, with no DevOps and no runtime lock-in.
You Don't Buy a Streaming Backend Anymore. You Provision One.
A streaming backend used to be a buy-or-build decision. With dozens of open-source media services on Open Source Cloud, you provision one in an afternoon and own every piece.