mcpfy cloud
The platform for everything after the code
Deploy from GitHub in minutes, watch every tool call with MCP-native analytics, and ship your server to any AI client. One pipeline covers the whole MCP lifecycle.
Our open source tools are used by developers at top companies
How it runs
One pipeline, end to end
- 1
Deploy
GitHub-native builds in seconds.
- 2
Test
Cross-client suites gate the merge.
- 3
Observe
MCP-native analytics and replay.
- 4
Publish
Checks and a submission pack.
Deploys that feel like web apps
Connect a repository once and every push ships. Watch each build move through cloning, analysis, build, deploy and health checks live — then promote, redeploy or roll back from the same table.
- mainProduction12s
- feat/chartsPreview9s
- fix/authPreview11s
➜ npx mcpfy deploy
✓ live → my-server.mcpfy.ai/mcp
A live endpoint for every branch
Every branch gets its own MCP URL, so a reviewer can point a real client at the change before it merges. Watch paths and branch allow lists keep monorepos from deploying on every commit.
➜ npx create-mcpfy-app
✓ tools/chart-sales.ts
✓ widgets/Chart.tsx
✓ mcpfy.config.ts
➜ mcpfy dev
ready on :4141 · 6 tools registered
tools/chart-sales.ts
export const chartSales = tool({
name: "chart_sales",
input: z.object({ period: z.string() }),
});Confidence before release
Suites describe scenarios, prompts and expected behaviour, then run your tools across clients and models with a model judge scoring every combination — so a regression in one client never reaches production.
San Francisco, CA
72°F
Partly cloudy · H 74° L 58°
→ tools/call
{
"name": "display_weather",
"arguments": { "city": "San Francisco" }
}← result20041ms
{ "content": [{ "type": "resource" }] }GPT-5.2
18/18
Claude
18/18
Gemini
17/18
Visibility after release
Every request through the gateway, refreshed in seconds: status, client, method and duration, with volume, latency and error-rate charts and optional payload capture when you need the full body.
last 24h ↑ 12.4%
By country34 regions
Tracesp50 42ms
- tools/call display_weatherclaude142ms
- resources/listchatgpt18ms
- tools/call search_docscursor842ms
- prompts/listcopilot12ms
- tools/call fetch_urlclaude503
p50
38ms
p95
112ms
errors
0.2%
Replay the whole conversation
Sessions group every interaction into one timeline — initialization, tool listing, calls and errors — classified by goal and outcome, filterable by client, country and tool activity, and deep-linkable from any analytics card.
- Protocol & discovery
- Tool conformance
- Security & policy
- !Domain / TLS / CSP
Submission pack
Capabilities
Everything Mcpfy Cloud handles for you
The full surface area of the platform, grouped the way you actually work through it.
Build & deploy
Creating servers
Start from a template, scaffold with the SDK, or connect a repository you already have.
Connect an existing server
Already hosted somewhere else? Point Mcpfy at the endpoint and keep the analytics, testing and publishing.
Deployments
Cloning, analysing, building, deploying and health checks — each step streamed live, with build and runtime logs per deploy.
Build & runtimes
Node and Python detected from your lockfile, or any Dockerfile. npm, pnpm, yarn and bun are picked up automatically.
Environment variables
Per-environment secrets, pasted straight from a .env file, scoped so previews never touch production credentials.
Domains
A free slug URL out of the box, custom domains by CNAME, and TLS certificates provisioned on verification.
Server settings
Override build and start commands, set watch paths for monorepos, restrict which branches deploy, and hold on CI.
Monitoring
Usage & traffic
Request volume, latency percentiles and error rate over time, with related failures grouped into single issues.
Sessions
Browse and replay whole conversations, with goals and outcomes classified for you and manual overrides when they are wrong.
Gateway requests
The raw event stream: status, session, client, method and duration, filterable and refreshed every few seconds.
Runtime logs
stdout and stderr streamed from the running deployment, searchable and scoped to the environment you are viewing.
Testing
Test suites
Scenarios and expected behaviour, executed across MCP clients and LLM models with pass or fail recorded per combination.
Model-judged results
A judge scores each response, so you catch the client- or model-specific regressions a status code would miss.
Run history
Every run kept, so you can see exactly which change broke which tool in which client.
Distribution
Connecting clients
One-click setup for Cursor, VS Code, Claude Desktop and Code, and the CLI agents — or copy the config for anything else.
Public chat
Turn the server into something anyone can try from a browser, with no MCP client installed.
Branding & SEO
Custom title, description, starter questions and an OG image, with a search preview before you share the link.
Publish
Publish checks
An audit of protocol compliance, initialization, tool schemas, error handling and metadata, scored by category.
End-to-end checks
Your server exercised live inside ChatGPT and Claude, confirming tools actually run and widgets actually render.
Guided fixes
Every failure comes with how to fix it — copy the prompt, or let the GitHub integration open the change for you.
Submission pack
Listing copy, per-tool justifications, reviewer test cases, screenshots and a demo video, generated for the directories.
Organisation
Team
Invite the people who ship with you and keep every server under one organisation.
Roles & permissions
Decide who can deploy, who can change secrets and who can only read the analytics.
GitHub integration
Install the app once. Webhooks handle the pushes, the checks and the branch previews from there.
API keys & authentication
Bearer-token API keys for the REST API, CLI login for your terminal, and OAuth for the servers that need it.
Billing & plans
Usage-based pricing with per-plan limits on projects, requests, retention and seats.
In the box
Built for MCP traffic, not generic HTTP
Builds
TypeScript, Python or a Dockerfile.
Previews
A live endpoint per branch.
Auth
OAuth 2.1, API keys and bearer tokens.
Scale
Multi-region, cold-start prevention.
Retention
Up to unlimited analytics history.
Rollback
Promote any past deploy in one click.
Payloads
Optional full request and response capture.
Notifications
Email on failed deploys, by default.
Ship your MCP server on Mcpfy Cloud
Explore the rest of the platform
Everything on Mcpfy Cloud shares one pipeline.