IN ORBIT AROUND ILINXA.COM

The backend for your apps.Run it from your AI.

Databases, file storage, content, forms and email, managed for you. Switch on the modules you need, invite your team, and let the AI you already use do the work.

MCPhttps://edge.ilinxa.cloud/mcp
One address for Claude, Cursor, VS Code and any MCP client.
  • 5MODULES
  • 68CURATED MCP TOOLS
  • 71CAPABILITIES
FIELD LIVE

01 / WHY EDGE

WHAT YOUR APP GETS

Everything behind your app. Nothing to run yourself.

edge is the backend your website or app talks to. We keep it running; you and your AI decide what it does.

  • Run it from your AI

    Connect Claude, Cursor, VS Code or any MCP client once, then ask in plain words. What you do in the dashboard, your agent can do too, inside the permissions you give it. A few owner-only steps stay with you.

    • Claude Code, Cursor, VS Code, Claude Desktop
    • Every action checked against your permissions
    • From your phone and claude.ai Coming soon
  • Managed for you

    A ready backend for your site or app. We run the servers and keep them updated and monitored, with encrypted backups on the schedule you set. Need more room as you grow? We raise your limits.

    • Encrypted, scheduled backups
    • Your site calls it with a scoped key
  • Your data, your choice

    Every project gets its own databases and file storage from day one. Prefer your own? Your email provider plugs in today; your own database and storage are on the way.

    • Built-in databases and storage
    • Your own email provider
    • Your own database Coming soon
    • Your own storage Coming soon
  • Modular, and growing

    Switch on only what a project needs: content, forms, email, project planning. A new module arrives every month.

    • Five modules today
    • A new one every month
  • Coming soon

    Module bundles

    A pack of modules set up for one job, switched on in one step. A company website, for example: content, a contact form and email, ready to fill.

  • Your team, not your password

    Invite people as members with their own sign-in instead of sharing one account. Each gets a role, and content editors can change only the pages you hand them.

    • Owner, admin and member roles
    • Invitations by email
  • For individuals and organizations

    Start with personal projects. Run a team as an organization, with shared projects, members and limits in one place.

    • Personal projects
    • Organizations with shared projects

02 / GET STARTED

THREE STEPS

From sign-up to your first request in three steps.

Create an account, connect the AI tool you already use, then ask it to build. No servers to set up.

  1. Create your account

    Sign up with email, Google or GitHub. Create your first project and switch on the modules it needs.

  2. Connect your AI

    Open your project's settings and find Agent connections. Pick your tool and a permission preset, and you get a token and a line to paste. Try it below.

    A

    Pick your tool

    Claude Code, Cursor, VS Code, Claude Desktop or the CLI on its own.

    B

    Pick a preset

    Read only, Editor or Everything. Each is a rule over the capability catalog.

    C

    Paste the setup line

    One-click add for Cursor and VS Code.

    $ claude mcp add --transport http ilinxa-edge https://edge.ilinxa.cloud/mcp --header "Authorization: Bearer nbat_..."

    Run this in the folder you work in. Add --scope user to make it available in every folder.

    PREFER A TERMINAL?Coming soon

    $ npm install -g @ilinxa/cli$ ilx auth login --browser

    Signs the CLI in through your browser, Google and GitHub accounts included.

  3. Ask it to do the work

    Your agent first reads what it is allowed to do, then does it. Steps that cannot be undone wait for your approval.

    Your AI
    • Add a contact form with name, email and message, and publish it.
    • Draft a page about our spring offer for me to review.
    • Show me this week's form submissions.
    • Done. The form is live and the page is waiting in drafts.
    Read the full guide

03 / MODULES

ENABLED PER PROJECT

Five modules today. A new one every month.

Switch a module on and it arrives complete: screens for your team, tools for your agent, an API for your app. No redeploy, and its database updates apply themselves on release.

  • cms

    Pages, collections, singletons, blueprints, versions, releases, sites, translation, redirects, webhooks, import and export, preview.

    • REST
    • MCP 16
    • CLI
    • UI
  • forms

    Builder, immutable versions, hosted page, public submissions, file fields, CSV, JSON and XLSX export, retention, notifications.

    • REST
    • MCP 13
    • CLI
    • UI
  • email

    Send log, templates, thank-you emails. Platform sender, or your own Resend or Amazon SES account.

    • REST
    • MCP 7
    • CLI
    • UI
  • pm

    List, board, timeline, cycles, custom roles, live updates.

    • REST
    • MCP 20
    • CLI
    • UI
  • tasks

    A small sample module to copy.

    • REST
    • MCP 3
    • CLI
    • UI
  • Every month

    A new module joins the list, ready to switch on in any project.

  • BundlesComing soon

    Modules packed for one job and switched on together.

04 / THREE WAYS IN

ONE CODE PATH

Click it, script it or ask for it. Same rules.

Update a page as an editor, from a script, or through an agent. The capability check, the rate limit and the audit row are one code path.

Editor / In the admin
Pages / About
HeadlineWe build the edge your team runs on.StatusDraftPublish

A form per content type, drafts, versions and releases. Members see what their role allows.

Script / From the terminal
ilx
$ ilx agent run --module cms \    --operation nodes.update \    --params '{"n":"<page id>"}' \    --body @headline.json --json{ "ok": true, "data": { ... } }

Any CMS operation through ilx agent run, with a --json envelope you can pipe.

Agent / Through MCP
Claude Code

"Update the About headline to match the new tagline."

cms.update_page . capability cms:page.write

ok . allowed by the Editor preset

Connect Claude or any MCP client. The agent gets the capabilities you listed, a rate limit, and nothing above the role of whoever connected it.

AUDIT LOG

WHOSURFACECAPABILITYRESULT
An editoradmincms:content.writeupdated About
ilx connection site-scriptCLIcms:content.writeupdated About
ClaudeMCPcms:page.writeupdated About

05 / ONE PERMISSION MODEL

71 CAPABILITIES

Everything a person can do, an agent can reach. Nothing more.

Scoped and capped

71 capabilities, each a read or a write with a role floor. A connection never exceeds the person who granted it.

Parity, checked

Every API route is either reachable by an agent or has a written reason it is not. A coverage gate in CI fails the build when one has neither.

Agents teach themselves

whoami, describe_capability, agent.list_operations and served module skills tell an agent what it may do before it tries.

06 / FOUNDATIONS

MANAGED FOR YOU

Built on proven parts.

RUNTIME

  • Next.js 16
  • Fastify 5
  • PostgreSQL 18
  • Redis 8 + BullMQ
  • SQLite per project
  • S3 media
  • Docker compose
  • Caddy or Traefik TLS
  • pg-backup
  • Grafana overlay
  • GlitchTip

SECURITY

  • argon2id
  • OAuth: Google + GitHub
  • Short-lived sessions
  • CSRF
  • Rate limits
  • SSRF guard with DNS pinning
  • KEK keyring
  • SHA-256 tokens
  • Encrypted project backups
  • A human-only tier no agent can hold

Backed by a written disaster-recovery runbook.

07 / YOUR FIRST PROJECT

THE ORBIT CLOSES HERE

Open a project. Connect your AI.

Create an account, switch on a module and connect your agent. From then on your team, your AI and your app all work on the same project.