The open-source MCP gateway for your whole company

Give every employee and agent the approved tools and skills they need, inside the AI clients they already use. One sign-in, one policy, one audit log.

Get startedBook a demo

What is an MCP gateway?

An MCP gateway sits between AI clients (Claude, ChatGPT, Cursor, your own agents) and the MCP servers that connect them to company tools. Instead of every person configuring servers and tokens on their laptop, the gateway gives everyone one URL, signs them in through your identity provider, and applies the same access rules and audit log to every tool call.

  1. 01

    MCP servers configured per laptop

    Each person wires up their own servers, so no two setups match.

  2. 02

    Tokens pasted into local config files

    Personal API keys end up on laptops with no owner and no expiry.

  3. 03

    Every new tool means another security review

    Without one control point, each integration is approved and logged separately. A gateway puts sign-in, access rules, and the audit log in one place.

How the Metorial MCP gateway works

Every AI client in your company reaches company tools through one governed path, on the identity of the person using it.

  1. Sign in once with SSO

    People sign in to their portal with your identity provider. Their groups decide which integrations and skills they see.

  2. Paste one URL into any client

    Each person adds their Magic MCP URL to Claude, ChatGPT, Cursor, Codex, or their own agent. The same approved tools show up everywhere.

  3. Every call goes through the gateway

    Each tool call is checked against policy, runs on that person's own connected account, and is recorded with their identity.

Connect your AI client

Every client uses the same Magic MCP URL. Each person copies theirs from their portal, adds it once, and signs in with SSO. No API keys go into config files.

  1. Open Customize, then Connectors. Select Add, then Custom, then Web.
  2. Name it Metorial, paste your Magic MCP URL (<your-magic-mcp-url>) and continue.
  3. Select Add and sign in to load your tools.

Tools appear based on the access your groups grant, and every call is logged with your identity.

What you get

Give every employee and agent the approved tools and skills they need, inside the AI clients they already use. One sign-in, one policy, one audit log.

  1. 01

    One URL behind your SSO

    People sign in with your identity provider, such as Okta or Entra, and existing groups carry over. Each role sees only the tools it is approved for.

  2. 02

    1,000+ integrations, plus your own

    Verified integrations work out of the box, versioned and with per-user OAuth. Bring custom, remote, or Docker-based MCP servers for internal systems.

  3. 03

    Shared skills

    Teams edit skills in the browser, sync them to Git, and deliver them by role, so a workflow written once reaches everyone who needs it.

  4. 04

    Every call on the record

    Each tool call is tied to the person behind it and recorded, so security can answer who did what without asking around.

  5. 05

    Works in the clients you already use

    Claude, ChatGPT, Cursor, Codex, or an agent you build: anything that speaks MCP connects through the same URL and reaches the same approved tools.

  6. 06

    Open source, self-host if you need to

    The core is open source on GitHub. Run it in your own cloud or on-prem, or use the hosted platform.

One link to every tool

Paste a single Magic MCP link into any client and the agent reaches 1000+ verified integrations, custom providers, and internal systems. No servers to wire up.

Agents act as the real person

People sign in with company SSO and connect through one link. Agents use that identity and can never exceed what the person behind them is allowed to do.

See every call underneath

Tracing records each tool call and provider response, so you can debug behavior and review what agents did across every connection from one place.

Customer stories

Crunched

“Metorial lets us turn customer connector requests into something we can actually ship. We can focus on building the AI analyst experience, while Metorial handles the OAuth, permissions, and MCP infrastructure underneath.”

Markus Skagemo
Markus SkagemoHead of AI, Crunched
Read Story

Solution

Metorial Platform

Discover moreBook a demo

Metorial connects AI agents to your tools, data, and providers, and keeps every connection governed, secure, and observable. One gateway, with access control, tracing, and isolated infrastructure underneath.

lumon-industries.metorial.com
The Metorial platform console showing providers, integrations, traces, and access policies.
  1. Foundation

    One access layer for every agent surface

    Give every team shared integrations, providers, policies, and tracing instead of a separate connector stack for each AI product.

  2. Control

    Manage providers and access centrally

    Register and operate providers from one place, with access control and identity built into how agents use tools.

  3. Trust

    Built for production security and visibility

    Run connections in isolated infrastructure, check risky traffic, and trace every call so teams can rely on what they ship.

Product

Magic MCP

Discover moreBook a demo

Magic MCP is a single URL you paste into your AI client. Sign in with the login you already use, and your agent reaches every integration and tool you allow it to use.

lumon-industries.metorial.com
A single Magic MCP URL connecting an AI agent to the integrations a person is allowed to use.
  1. One URL

    Paste one link into any agent

    Drop a single Magic MCP URL into any MCP-compatible client. Nothing to install and no integration to wire up tool by tool.

  2. Every tool

    1000+ integrations, plus your own

    The same link reaches a growing catalog of verified integrations and the custom or remote MCP servers only your team has, kept current for you.

  3. Your access

    Only what you allow, never more

    Sign in with the identity you already use and stay tokenless. Behind the link, every call runs on the real user and reaches only the tools they’re allowed.

Product

Integrations

Discover moreBook a demo

Start from 1000+ verified integrations or add your own, and give every one a governed path to your agents under existing SSO and access policies.

lumon-industries.metorial.com
The Metorial integrations catalog showing verified providers, tools, and access controls.
  1. Connect

    Any tool, ours or yours

    Start with 1000+ verified integrations or add the internal systems and private APIs only your team has as custom, remote, or Docker-based providers.

  2. Reach

    One link for every approved tool

    Agents reach the integrations they’re allowed to use through a single connection, with no connector project per tool.

  3. Govern

    Scoped and traceable by default

    Access follows the groups and policies you already manage, down to individual tools and actions, and every call stays on the record.

Integrations

Slack

Slack: connect with bot OAuth or user OAuth to work with messages, conversations, search, files, Canvases, Lists, profiles, reactions, reminders, and workspace productivity settings.

GitHub

Manage repositories, issues, and pull requests. Create and configure branches, star repositories, review code, and merge changes. Automate CI/CD workflows with GitHub Actions, manage workflow runs, secrets, and artifacts. Track issues with labels, milestones, and assignees. Search across code, repositories, issues, and users. Manage organizations, teams, and memberships. Create and manage projects, gists, packages, deployments, and environments. Access security alerts including code scanning, secret scanning, and Dependabot alerts. Read and write file contents in repositories. Manage webhooks, notifications, and codespaces.

Hubspot

Create, read, update, and delete CRM objects including contacts, companies, deals, tickets, and custom objects. Manage associations between CRM records. Track engagements such as calls, emails, meetings, notes, and tasks. Send transactional and marketing emails, manage campaigns, and handle subscription preferences. Build and manage marketing workflows, sequences, and automation actions. Manage CMS content including blog posts, landing pages, and HubDB tables. Handle commerce data including products, orders, invoices, quotes, and payments. Create and manage contact lists for segmentation. Configure deal, ticket, and order pipelines with custom stages. Search and filter CRM records. Bulk import and export CRM data. Manage users, teams, and account settings.

Linear

Create, read, update, and delete issues across teams with support for priorities, labels, assignees, due dates, and workflow states. Manage projects that group related issues, and organize work into time-boxed cycles (sprints). Create and manage documents, comments, and file attachments. Search issues using vector similarity or rich filters. Configure teams, workflow states, and labels. Track customers and link them to issues.

Salesforce

Manage CRM data including Accounts, Contacts, Leads, Opportunities, Cases, and custom objects. Create, read, update, and delete records. Query data using SOQL and search across objects using SOSL. Perform bulk data operations for large-scale imports, exports, and migrations. Execute composite requests to batch multiple operations in a single API call. Access analytics, reports, and dashboards. Manage files and attachments associated with records. Interact with Chatter feeds, posts, and groups for social collaboration. Manage org metadata including custom objects, fields, layouts, and workflows. Query data using GraphQL for precise data retrieval across related objects.

Microsoft Teams

Send, read, edit, and delete messages in Microsoft Teams channels and chats, with @mentions. Create and manage teams, channels, members, and tags. Schedule and look up online meetings, read user presence, and manage Shifts schedules. Connect a Teams bot to receive and reply to messages in chats and channels.

How Metorial compares to other MCP gateways

The main differences are who can run it, what comes in the catalog, and whether teams can share skills. For ten gateways in more depth, see the best MCP gateways in 2026.

FeatureMetorialRunlayerMintive (formerly MintMCP)Cloudflare MCP portalsBuild in-house
Open sourceOpen core on GitHub, self-host or cloudNoNoNoYour own code
Integration catalog1,000+ verified integrationsScans 18,000+ public MCP servers for admins to approve100+ hosted integrationsNone, bring your own serversYou build each one
SSO and per-role tool accessYes, SAML SSO on the Enterprise planYesYes, SSO and SCIM listed as enterprise optionsAccess policies per server, tool allowlists per portalYou build it
Shared skillsEdited in the browser, synced to GitPushed from a Git repo with a CLIOnly inside Coworker Agents (beta)NoneYou build it
Works in Claude, ChatGPT and CursorYesYesYesClaude documented, ChatGPT and Cursor not documentedDepends on what you build
Public pricingYes, free Dev plan, Scale at $250 a monthNo, book a demoNo, custom pricingYes, within Cloudflare Zero Trust plansEngineering time

Checked against each vendor's own site and docs on October 16, 2026.

Give your whole company one MCP gateway

One URL, behind your SSO, with the tools and skills each team is approved for. See how it works.

Book a demoStart free

Frequently asked questions

Common questions about MCP Gateway.

  1. What is an MCP gateway?
    A single control point between your AI clients and every MCP server they call. It handles sign-in, access control, and logging once, centrally, instead of once per tool and per client. You need one when more than one team or more than one AI client needs the same integrations.
  2. An MCP server exposes the tools of one app or system. A gateway sits in front of many servers and decides who may call which tool, runs the sign-in, and records each call, so those rules are not rebuilt inside every server.
  3. They solve the same class of problem at different layers. An API gateway fronts HTTP endpoints and routes by path and method. An MCP gateway fronts tools that a model chooses at runtime, so it also has to handle per-user credentials and record which identity a call ran as.
  4. Shadow MCP is MCP servers people set up on their own machines, with personal tokens and no record of what they call. A gateway replaces those setups with one approved URL behind SSO, so tool use is visible and follows company policy. See what is shadow AI.
  5. Any MCP-compatible client: Claude, ChatGPT, Cursor, Codex, or your own agent. Each connects through one Magic MCP link and reaches the same approved tools.
  6. Yes. The core is open source on GitHub and can run on your own infrastructure or on-prem. The hosted platform is there if you would rather not operate it.
  7. People sign in with company SSO and existing groups carry over. Policies and access control decide which tools each group can call, and an agent never exceeds the access of the person behind it.
  8. Yes. Register custom, remote, or Docker-based servers you already run, and they get the same sign-in, policy, and logging as the 1000+ verified integrations. You do not rewrite a server to put it behind the gateway.
  9. It adds one network hop plus policy evaluation, typically a few milliseconds. That is small next to the model inference and the upstream API call in the same request, and it removes the per-client credential lookups a direct setup repeats.
  10. Usually not. One developer with one client and a few servers is better off configuring them directly. The gateway earns its place when a second client or a second person needs the same integrations, because that is when per-person setup and shared credentials start.