MCP server: connect AI assistants to your organization

Let Claude, ChatGPT, or any MCP-compatible assistant work with your membership data directly — reading records, answering questions, and taking actions you've explicitly allowed.

  • One connection covers the whole platform: members, events, payments, contracts, communications, governance, and learning
  • Scoped API tokens control exactly what an assistant can see and do, area by area
  • Every token is revocable in one click, with usage tracking and an audit trail

Built on the Model Context Protocol, the open standard for connecting AI assistants to software.

The MCP connection at a glance

Protocol Model Context Protocol (MCP) — works with Claude, ChatGPT, and any MCP-compatible client
Coverage The full Orgo API: 458 endpoints across 118 resource families, from member records to event tickets to invoices
Operations Read and write — assistants can look up data, answer questions, and make changes the token allows
Access control Per-area None / Read / Write scopes across four categories: Members, Events, Financial, and Communications
Authentication A scoped API token, generated in the admin UI; the secret is shown once and never retrievable again
Auditability Last-used tracking per token, instant revoke, and an audit log of every scope change
Data isolation Every call is scoped to your organization's own workspace; cross-organization access is blocked at the database layer

The MCP server is for

Admins who already use AI assistants

Ask Claude or ChatGPT about your own members, events, and payments instead of exporting spreadsheets and pasting data into a chat.

Teams automating routine work

Let an assistant draft the follow-up list, register the attendee, or update the record — within the exact permissions you granted.

Developers and integrators

One standard connection instead of custom glue code: the assistant discovers the API by itself, endpoint by endpoint.

Everything the MCP server gives your organization

Works with your AI assistant

Orgo's MCP server speaks the Model Context Protocol, the open standard AI assistants use to connect to external software, so it is not tied to a single vendor.

Connect from any assistant that supports MCP:

  • Claude (claude.ai and Claude Code)
  • ChatGPT and other MCP-compatible assistants
  • Custom agents and internal tools your team builds

Instead of a fixed list of pre-wired commands, the assistant explores the API itself: it lists the available operations, reads how each one works, and then calls the one it needs. New Orgo features become available to assistants without reconfiguring anything.

The server ships its own docs as MCP resources, so the assistant can read how authentication, pagination, and filtering work before making a single call.

Full platform coverage

The MCP server is a gateway over the entire Orgo API, not a curated subset, so anything your organization manages in Orgo is reachable by an assistant with the right permissions.

A single connection covers:

  • People — members, contacts, families, companies
  • Events — tickets, attendance, speakers, vouchers
  • Payments — invoices, products, billing
  • Contracts and e-signatures
  • Communications — newsletters, discussions, notifications
  • Governance and learning — voting, org charts, courses, badges

Assistants are not limited to lookups. With a write-scoped token, an assistant can create and update records — register an attendee, update a member's details, or draft a newsletter — using the same validated API the Orgo apps use.

Scoped access tokens

Access is granted through API tokens with a permissions matrix, so an assistant only ever gets the capabilities you explicitly chose.

Each token sets None, Read, or Write independently for four areas:

  • Members — records, groups, courses, contracts, badges, votes
  • Events — events, attendance, tickets, speakers
  • Financial — invoices, payments, products, billing
  • Communications — newsletters, discussions, posts

A token for an event-planning assistant can write Events data while seeing nothing financial at all.

A full-access read-only token lets an assistant answer questions across the whole platform while rejecting every modification attempt — the safest starting point for most organizations.

Every endpoint in the platform is classified into a permission category, and anything unclassified is denied automatically. Sensitive areas — authentication flows, billing configuration, webhooks administration, data imports — are permanently off-limits to scoped tokens, regardless of what they were granted.

Security & auditability

Tokens are managed by your organization's admins, and every step of a token's life is visible and reversible.

Only tenant admins can create tokens, and each token is:

  • Shown once at creation, never retrievable again
  • Optionally set to expire on a date you choose
  • Revocable instantly, cutting off any integration that used it

Each token records when it was last used, and every change to a token's permissions is written to an audit log, so you can always answer who could access what, and when.

Every call is scoped to your organization's own workspace. Cross-organization reads are blocked at the persistence layer, so even a misdirected query cannot leak another organization's data.

Built on the Orgo API

The MCP server sits on the same REST API that powers the Orgo apps themselves, so assistants get first-class, validated access rather than a bolted-on integration.

Every resource follows the same conventions — pagination, filtering, ordering, validation — so once an assistant learns one endpoint, it knows the shape of all of them.

The same developer platform also offers:

  • Webhooks for pushing events (payments, registrations, contract signatures) to your own systems
  • OAuth 2.0 for building "Log in with Orgo" into third-party apps
  • Plain REST access for scripts and scheduled jobs

“With an organization of 8,000 members and more than 80 chapters coordinated almost exclusively by volunteers, Orgo is at the heart of our digitisation process, allowing us to focus on our mission - learning by doing, nature, and personal development.”

Photo of Andrei Avram

Andrei Avram

President, Romanian Scouts

MCP server FAQ

An MCP server is a connection point that lets AI assistants interact with a piece of software through the Model Context Protocol, an open standard. Instead of copying data into a chat window, the assistant connects to the software directly, reads the data it is allowed to see, and performs actions it is allowed to take.

Any assistant that supports the Model Context Protocol, including Claude and ChatGPT, as well as custom agents your team builds. The protocol is an open standard, so the connection is not tied to one vendor.

That is entirely up to the token you create. A read-only token lets an assistant answer questions while rejecting every modification. A write-scoped token additionally allows changes, but only in the areas you granted — for example, Events but not Financial.

Each API token carries its own permissions matrix: None, Read, or Write, set independently for Members, Events, Financial, and Communications data. Enforcement is default-deny, meaning anything not explicitly granted is refused, and sensitive areas like billing configuration and authentication settings are off-limits to tokens entirely.

An admin generates an API token in Orgo's developer settings, chooses its permission scopes, and adds the token to the assistant's MCP connection settings. The token secret is displayed once at creation, so it should be stored securely at that moment.

Revoke it. Revocation takes effect immediately and any integration using the token stops working. Tokens also record when they were last used, which helps identify whether an exposed token was ever exercised.

Yes. Every API call is scoped to your organization's own workspace, and cross-organization reads are blocked at the database layer. See our data privacy page for details on hosting, encryption, and GDPR compliance.

No. Creating a token and pasting it into an assistant's connection settings is an admin task, not a coding one. Developers become useful when you want deeper automation — custom agents, webhooks, or scripts on top of the same API.