GitHub to HTTP endpoint in 30 seconds
Deploy any MCP server from GitHub as an HTTP endpoint. Works with n8n, Dify, Voiceflow, Langflow, and any platform that needs MCP over HTTP. Open source, self-hostable.
Deployed Servers
12
From GitHub repos
Auto-Deploys
47
On git push this month
Avg Deploy Time
28s
GitHub to HTTP
Recent Deployments
Auto-deploy enabledmodelcontextprotocol/server-github
main · a3f8c12
anthropics/mcp-server-fetch
main · b7d2e45
acme/internal-mcp-server
main · c9e1f33
Getting MCP servers to HTTP
shouldn't be this painful
Workflow automation platforms need HTTP endpoints. Most MCP servers are stdio-only. DeployStack bridges the gap — GitHub repo to HTTP URL in 30 seconds.
Your platform needs HTTP. The server only speaks stdio.
n8n, Dify, Voiceflow, and Langflow all need HTTP endpoints for MCP servers. But most MCP servers on GitHub only support stdio. You're stuck bridging the gap yourself.
Hosting one MCP server takes hours.
Write a Dockerfile, configure a reverse proxy, set up SSL, deploy to Fly.io or Railway. Repeat for every MCP server. Each one is another piece of infrastructure to maintain.
npx mcp-remote... and then what?
mcp-remote requires Node.js installed locally, runs as a local proxy, and doesn't give you a persistent HTTP endpoint. It's a workaround, not a solution.
Every update means redeploying manually.
The MCP server author pushes an update. You have to notice, pull changes, rebuild, redeploy. No auto-deploy, no webhook, no git push integration.
Getting MCP servers to HTTP
shouldn't be this painful
Workflow automation platforms need HTTP endpoints. Most MCP servers are stdio-only. DeployStack bridges the gap — GitHub repo to HTTP URL in 30 seconds.
No Docker. No mcp-remote. No DevOps.
Your platform needs HTTP. The server only speaks stdio.
n8n, Dify, Voiceflow, and Langflow all need HTTP endpoints for MCP servers. But most MCP servers on GitHub only support stdio. You're stuck bridging the gap yourself.
Hosting one MCP server takes hours.
Write a Dockerfile, configure a reverse proxy, set up SSL, deploy to Fly.io or Railway. Repeat for every MCP server. Each one is another piece of infrastructure to maintain.
npx mcp-remote... and then what?
mcp-remote requires Node.js installed locally, runs as a local proxy, and doesn't give you a persistent HTTP endpoint. It's a workaround, not a solution.
Every update means redeploying manually.
The MCP server author pushes an update. You have to notice, pull changes, rebuild, redeploy. No auto-deploy, no webhook, no git push integration.
From GitHub to HTTP
in four simple steps
No Dockerfiles, no reverse proxies, no infrastructure. Just point, deploy, and use.
Connect your GitHub repo.
Authenticate with GitHub, then select the repository containing your MCP server. DeployStack validates the package.json and checks for the MCP SDK dependency automatically.
Deploy in 30 seconds.
DeployStack downloads the repo, runs npm install and npm build, resolves the entry point, and spawns your MCP server. You get a live HTTP endpoint instantly.
Paste the URL into your platform.
Copy the HTTP endpoint and paste it into n8n's MCP Client node, Dify's MCP config, Voiceflow's MCP tool, or any AI client that supports MCP over HTTP/SSE.
Auto-deploy on every git push.
Enable webhooks and DeployStack redeploys automatically when you push to your branch. Your MCP server stays up to date without manual intervention.
Connect your GitHub repo.
Authenticate with GitHub, then select the repository containing your MCP server. DeployStack validates the package.json and checks for the MCP SDK dependency automatically.
Deploy in 30 seconds.
DeployStack downloads the repo, runs npm install and npm build, resolves the entry point, and spawns your MCP server. You get a live HTTP endpoint instantly.
Paste the URL into your platform.
Copy the HTTP endpoint and paste it into n8n's MCP Client node, Dify's MCP config, Voiceflow's MCP tool, or any AI client that supports MCP over HTTP/SSE.
Auto-deploy on every git push.
Enable webhooks and DeployStack redeploys automatically when you push to your branch. Your MCP server stays up to date without manual intervention.
From GitHub to HTTP
in four simple steps
No Dockerfiles, no reverse proxies, no infrastructure. Just point, deploy, and use.
Works with any MCP server that uses the official SDK.
Built for workflow automation
These platforms need HTTP endpoints for MCP servers. DeployStack gives them exactly that.
n8n
MCP Client node needs HTTP URLs. Paste your DeployStack endpoint directly into the MCP Client node configuration.
- HTTP endpoint compatible
- Works with MCP Client node
- Auto-deploy keeps it current
Dify
Dify v1.6+ only supports HTTP/SSE for MCP. DeployStack hosts your stdio servers as HTTP endpoints that Dify can connect to.
- HTTP/SSE endpoint
- Works with Dify v1.6+
- No local server needed
Voiceflow
Voiceflow explicitly says local MCP servers are not supported. DeployStack gives you the hosted endpoint Voiceflow requires.
- Remote HTTP endpoint
- No local MCP needed
- Production-ready hosting
Langflow
Langflow supports HTTP/SSE connections for MCP in production flows. Deploy your servers to DeployStack for reliable remote access.
- HTTP/SSE compatible
- Production-grade uptime
- Works with any flow
Everything you need for MCP deployment
Deploy your first MCP server in 30 seconds
GitHub repo to HTTP endpoint. No Docker, no DevOps, no mcp-remote. Open source and free to start.