# Run your LMS from Claude, ChatGPT or Gemini (Cubite MCP server)

In short: Cubite ships its own hosted MCP server. It exposes 137 administration tools covering courses, learners, enrollments, certificates, grading, discussions, reports and site settings. Claude and ChatGPT connect with a browser sign-in (OAuth 2.1, no API key to copy), one connection can cover several of your sites, the assistant only sees the tools your connection has permission for, and it is included with every Cubite site.

- Endpoint: https://cubite.io/api/mcp/mcp (streamable HTTP)
- Auth: OAuth 2.1 with dynamic client registration; metadata at https://cubite.io/.well-known/oauth-authorization-server. API keys exist for command-line and server-to-server clients that cannot open a browser.
- Works with: Claude (web, desktop, Claude Code), ChatGPT in developer mode, Gemini CLI, Cursor, and custom agents built on the Anthropic or OpenAI APIs.
- Cost: included with your Cubite site ($290/mo or $2,900/yr, unlimited learners and courses). You bring your own assistant subscription.
- Page: https://cubite.io/mcp

## What the assistant can do

- Courses and content: create courses, write lessons and units, add quizzes, assignments and graded activities, upload SCORM or xAPI packages.
- Learners: invite and create users, enroll individuals or a whole cohort, manage groups, revoke access.
- Analytics and insights: site overview, per-course engagement, drop-off, broken exam questions, at-risk learners, grading turnaround, sales revenue.
- SEO and Search Console: for sites with Google Search Console connected, performance totals and trend, top queries and pages, ranked SEO opportunities.
- Learning paths: build multi-course journeys, order steps, set prerequisites, enroll learners into the path.
- Site and marketing: update settings and theme, rebuild the homepage, publish pages and blog posts.
- Search and research: search across courses, pages, posts and learning paths and read any of them in full; this also powers ChatGPT's research modes.

## Six ready-made workflows

In Claude they appear as slash commands. Each one shows you the plan before it writes anything.

| Workflow | What it does | Needs |
| --- | --- | --- |
| Write and Publish an SEO Post | Checks the blog for a near duplicate, drafts the post, links a real course, stages it hidden for review | Pages |
| Build a Course From an Outline | Turns a syllabus, SOP or transcript into sections, units and graded checks, then fills in catalog metadata | Courses |
| Onboard a Learner Cohort | Dedupes a learner list against existing enrollments, picks a group or a bulk enroll, hands back sign-in links | Enrollments |
| Assemble a Learning Path | Orders existing courses into a guided path with unlock rules and prerequisites | Learning paths |
| Audit Catalog and Enrollments | Read-only report: published courses with no lessons, missing outlines, missing SEO, stalled enrollments | Courses (read) |
| Refresh Homepage and Site SEO | Rebuilds the homepage from courses that actually exist, keeping real testimonials and stats | Site |

## Connect your assistant

Claude and ChatGPT sign in through your browser, so there is no key to copy.

- Claude: use the Add to Claude link on https://cubite.io/mcp, or open Settings, then Connectors, then Add custom connector and paste the endpoint URL. Claude opens a Cubite sign-in; tick the site or sites to cover; approve the permissions. The tools appear in your next chat.
- Claude Code: `claude mcp add --transport http cubite https://cubite.io/api/mcp/mcp`, then run `/mcp` to complete the browser sign-in on first use.
- ChatGPT: turn on developer mode in settings, add a connector, paste the endpoint URL as the server address, sign in, tick the sites, approve the permissions. ChatGPT's research modes use the server's built-in search and fetch tools.
- Gemini CLI and other HTTP clients: point the client at the endpoint; use an API key with the scopes you need where a browser sign-in is not possible.

## Permissions

Granted per connection; the OAuth screen lists exactly what is requested before you approve. A write permission also grants the matching read (except forms:write, which manages templates only and never reads learner responses).

pages, courses, enrollments, users, members, groups, learning-paths, question-banks, packages (xAPI and SCORM), site, media, reports (aggregate analytics, never learner names), search-console, certificates, discussions, grading, forms. Each has read and, where it applies, write.

## Security and data handling

- A connection is bound to exactly the sites you tick on the sign-in screen; the server validates every request against that approved set, so an assistant can never reach another tenant.
- Disconnect from Admin, then any connected site, then Integrations; revoking there immediately invalidates the credential for every site it covered.

## Questions

- What is MCP? The Model Context Protocol is an open standard for connecting AI assistants to external systems. An MCP server exposes tools an assistant can call.
- Do I need to paste an API key into Claude or ChatGPT? No. Those clients use OAuth in your browser.
- Does this cost extra? No. The MCP server is included with your Cubite site.
- Other LMS vendors ship MCP servers too. Cubite's is broad (137 tools, full administration) and self-serve; it does not claim to be the first.

Talk to a person: https://cubite.io/go/book?src=mcp-md or hello@cubite.io.