MCP
mcp · setup
Connect Cursor, Claude Code, or Codex with @luno-cms/mcp — the agent interface for schema through publish.
- $npx @luno-cms/mcp setup
- →1) Claude Code 2) Cursor 3) Codex
- ✓Wrote .cursor/mcp.json + skills
- ✓MCP server luno-stg ready
LUNO is a hosted AI-era Backend Platform.
Build, operate, and govern production backends with AI agents
Let AI agents build and operate production backends. Humans keep control of what ships.
Why now
AI agents are not only here to write code.
They read data, operate services, execute changes, and finish work.
That shift needs a production backend agents can actually operate — with humans still governing what ships.
Your code editor
import { createClient } from '@luno-cms/mcp';
export async function initBackend() {
return createClient({ env: 'production' });
}TERMINAL
- $ npx @luno-cms/mcp setup
- Step 1/3 · Detecting Cursor / Claude Code / Codex
- Step 2/3 · Writing MCP config + skills
- Step 3/3 · Connecting to LUNO staging
- Success! LUNO MCP server ready
The new backend
Production backends built for AI agents
Human-first backends were not designed for agents to build, change, and operate. LUNO is agent-readable from schema to publish — with human governance on every production boundary.
Operating Model
AI agents run the backend
Agents build and change backends through MCP and APIs. Humans review and approve what reaches production.
BUILDAI builds the backendHand over requirements. Agents create schemas, forms, content structures, and other backend resources. OPERATEAI runs the backendThrough MCP or APIs, agents read, create, update, revise, and publish backend resources. GOVERNHumans control productionPermissions, review, approval, and audit — humans decide what agents can change and what ships.
For AI Agents
Your agents. Your tools.
Connect from Claude Code, Cursor, Codex, and the agents you already use. MCP, API, and CLI are agent interfaces — not the product.
live
Cursor
Claude
Claude Code
Codex
OpenAI
Copilot
GitHub
For Humans
AI runs it. Humans control it.
People and agents share the same production backend. Agents execute; humans govern changes and publishing — Console stays approachable when you need a dashboard.
Edit and publish content in place
Entries, fields, and publish flow — a human admin surface you can use as-is.
GOVERN
Keep humans in control of production
AI agents can build and operate the backend. Humans decide what they can change and what reaches production.
ContactProduction governanceScoped keys · Approval · Audit
- People
- Agents
- 01Scoped agent keys
Agents work only inside scoped keys and permissions — authority is bounded by design.
- 02Human approval
Approval and publish boundaries stay with the org. Review before changes go live.
- 03Audit trail
See who published what. Agent actions land on the same operational trail as humans.
Production Backend
Let AI agents run your backend.
Hand your AI coding agents a production backend. Content, forms, auth, storage, APIs, and publishing are already built into the platform — no backend to assemble from scratch.
See operator surfacesInfrastructure
Stop building the infrastructure.
The servers, runtime, database, mail, and secure access are already in place.
Learn moreIdentity & Access
Stop rebuilding identity.
Authentication, authorization, MFA, SSO, and organizations are ready from the start.
Learn moreContent & Data
Stop rebuilding content systems.
CMS, forms, data, storage, and search are already part of the backend.
Learn moreApplication
Stop wiring backend services together.
APIs, webhooks, workflows, jobs, and notifications work together out of the box.
Learn moreSecurity
Stop adding security afterward.
Access control, audit logs, encryption, secrets, and policies are built into the foundation.
Learn moreDelivery & Operations
Stop maintaining the backend.
Deployment, environments, monitoring, logging, and backups are handled by the platform.
Learn moreHosted & Secure
Trust the infrastructure underneath
Firewall, updates, and edge protection keep running. Security is proof you can ship production — not the story.
live
- TrafficProduction load
- FirewallBlocked at the edge
- UpdatesPatches applied for you
Proof
Built for real agent workloads.
142 → 11 tool calls
Same backend task with improved MCP schema — 142 calls and ~236 minutes down to 11 calls and 12.8 minutes.
12 → 0 errors
Agent operations on a structured backend surface — fewer round trips, fewer failures, production-ready results.
Human governance
Scoped keys, approval, and audit — agents operate fast; humans keep production under control.
Let AI agents run your backend.
Keep production under human control. Start building with MCP, API, and Console on one platform.
npx @luno-cms/mcp setup
Contact