Claude now provides identity for MCP connectors itself: 14 integrations without per-user OAuth

Anyone who has rolled out MCP connectors across a team knows the pain. You connect Claude to a task tracker, cloud storage and a CRM, and then every employee has to click “Authorize,” enter passwords and figure out why a colleague’s token works while theirs does not. At companies with a proper identity provider—an IdP such as Okta or Entra—single sign-on solves the problem. Without an IdP, the manual busywork remains.

Judging by a short post on X from @automater_ai, Anthropic has introduced a new feature for precisely this situation.

What changed

Claude can now act as the identity provider for MCP connectors itself. This applies to organizations on Team and Enterprise plans that do not have their own IdP. Instead of every user completing OAuth authorization separately for every service, an administrator configures access once, while Claude verifies each user’s identity.

A simple example: a 30-person team connects 5 connectors. Previously, that meant as many as 150 separate authorizations, some of which would inevitably break a week later. Now, if the setup works as described, the organization configures it once.

The feature is in beta and supports 14 connectors. The post does not provide a complete list, so check whether your services are included.

The main caveat: no MFA verification

The post’s author specifically notes that this setup has no MFA verification. For security teams, that is the key line. If access to a connector previously required a person to verify their identity with a second factor on the service side, Claude now verifies the identity itself, and that additional barrier may disappear.

In practice, this means that compromising a single Claude account could potentially provide access to every connected connector. Through them, the agent may be able to read documents, write to tickets and perform actions on the user’s behalf. The broader the connectors’ permissions, the costlier a mistake becomes.

What to do in practice

• If you already have an IdP, you probably do not need this feature. Single sign-on with MFA is more robust.
• If you do not have an IdP and your team is small, the feature can save hours of setup. Start with read-only connectors, however.
• Enable MFA on the Claude account itself, since this verification is absent further down the chain.
• Do not use this setup to connect payment systems, production databases or admin panels until the beta has been thoroughly tested.
• Check Anthropic’s documentation. The post on X received only 40 views; it is a second-hand account, not a primary source.

The market behind all of this

The same post cites a Menlo estimate: the coding-agent market is worth $34 billion, with Claude Code and Codex accounting for $24 billion. If the figure is correct, the two products control about 70% of the market. That explains why Anthropic is simplifying enterprise adoption. The less friction involved in connecting integrations, the easier it is for a vendor to sell subscriptions to entire teams rather than individual enthusiasts.

It is worth remembering that this is an analyst estimate, not audited revenue. The post does not disclose the calculation methodology.

Limitations of this report

There is only one source: a short post from a small account. It offers no details on exactly how tokens are issued, how access revocation works, how activity is logged or what permissions the agent receives. For an adoption decision, these questions matter more than the announcement itself.

Conclusion

For small teams without an IdP, this is a genuine quality-of-life improvement: fewer manual authorizations and faster onboarding. But the absence of MFA turns convenience into a deliberate trade-off. Enable the beta for a limited set of connectors, monitor access logs and wait for official documentation before rolling it out across the entire company.