惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

美团技术团队
Microsoft Azure Blog
Microsoft Azure Blog
freeCodeCamp Programming Tutorials: Python, JavaScript, Git & More
B
Blog
Y
Y Combinator Blog
博客园_首页
有赞技术团队
有赞技术团队
博客园 - Franky
腾讯CDC
G
Google Developers Blog
Recent Announcements
Recent Announcements
博客园 - 【当耐特】
D
Docker
The GitHub Blog
The GitHub Blog
MyScale Blog
MyScale Blog
H
Help Net Security
Apple Machine Learning Research
Apple Machine Learning Research
A
About on SuperTechFans
D
DataBreaches.Net
T
The Blog of Author Tim Ferriss
V
V2EX
U
Unit 42
aimingoo的专栏
aimingoo的专栏
WordPress大学
WordPress大学

The New Stack | DevOps, Open Source, and Cloud Native News

Agentic development hinges on verification. For cloud-native software, that is a runtime problem. AI agents need infrastructure: Why Europe’s regional cloud strategy matters Transform your AI coding agent into a deterministic Java Spring expert WeAreDevelopers is coming to the US to give unsung developers a bigger voice Cleaner AI training data, fewer bugs: Sonar’s SonarSweep explained Observability overload is drowning engineers Google’s DiffusionGemma is 4x faster than its other Gemma models Fable 5: Guardrails and burn rate are annoying users, who say it’s still better than Opus 4.8 The Anthropic leader who built Claude Code says he ditched prompting — now he just writes loops. AWS can now mathematically prove your VMs are isolated Microsoft pulled 73 GitHub repos after malware attack — but still won’t say who’s compromised Databricks wants to kill the “email me a file” problem for AI agent skills Ramp bets forward deployed engineers can do what off-the-shelf finance AI can’t Git real: AI agents aren’t just for solo developers anymore Anthropic launches Claude Mythos/Fable 5, but you better try it soon This AI agent startup ditched Anthropic for DeepSeek — and says it’s saving millions When your data model is the bottleneck: lessons from Medium’s feature store How long before we stop reading the code? The tokenmaxxing party is over, and Revenium is mopping up How AI is solving the memory crunch it created Microsoft’s pitch to enterprises: Ditch Azure Repos for GitHub, despite its rocky reliability record Claude Code’s biggest upgrade yet ran 5 agents at once — here’s what happened Why Anthropic just doubled Claude Cowork limits at no charge For years, Apache Cassandra handed this work to your team — 6.0 takes it back “A dangerous combination”: The 2 factors that can “corrupt” AI agent workflows With Foundry, Microsoft bets the enterprise AI battle is about reliability, not capability Microsoft unlocks Visual Studio for developers left behind by its own AI AI teams now deploy 1,000 times a month. Your pipeline wasn’t built for that. Microsoft just made the agent runtime free — and kept everything around it “Whoever builds the most joyous product wins”: The agent war begins
MCP gets its missing enterprise authorization layer
Frederic Lardinois · 2026-06-19 · via The New Stack | DevOps, Open Source, and Cloud Native News

Every enterprise company is seemingly trying to adopt the Model Context Protocol (MCP) to connect its AI agents to tools. But so far, authorizing those connections has meant employees clicking through an OAuth prompt for every server. For a while now, the MCP project has been working on the “Enterprise-Managed Authorization” extension, with the goal of allowing enterprises to control MCP server access centrally through their existing identity provider.

This extension is now stable and Anthropic and Microsoft are among the first to support it in their clients, including Claude, Claude Code, Claude Cowork, and Visual Studio Code, with Okta as the first identity provider.

After its launch, MCP quickly grew organically, but as is so often the case, the original spec wasn’t designed for enterprise use cases. The standard MCP authorization model was built for individuals. Authorizing servers, even today, means connecting service after service by hand, but this also means security teams can’t enforce consistent policy or keep a single audit trail — and there is always the risk of an employee connecting a personal account to a work tool.

“Logging in once and automatically having all your MCP connectors automatically set up is pretty magical,” says Tom Moor, the Head of Engineering at Linear, in today’s announcement.

The token handoff

Enterprise-Managed Authorization makes the identity provider the decision-maker for which servers a client can reach. An administrator sets the policy once, and employees sign in with the corporate identity they already use.

Unlike with OAuth, the exchange runs without a consent screen. During single sign-on, the client obtains a signed assertion from the identity provider that vouches for both the user and the application requesting access. It then presents that assertion to the MCP server’s own authorization server, which returns a scoped access token the client can then use to make its calls.

That assertion is actually an emerging OAuth extension called the Identity Assertion JWT Authorization Grant, or ID-JAG, now an IETF draft. Okta’s branded version is Cross App Access. Because ID-JAG is an open standard, any identity provider could implement it, though Okta is the only one to have shipped support so far.

Giving control back to IT

Credit: MCP

For an IT team, this ideally means an end to the inevitable sprawl of individual approvals. Now, an admin can enable a server for the organization (or specific teams or even individuals), and employees and their agents inherit access to it, scoped to the groups and roles they already hold.

Control and audit move into the identity provider’s console. Access decisions leave one trail across every connector, and revocation runs the same path as everything else, so deactivating a user cuts their MCP access at the same time.

Since corporate IT now controls this connection, mixing and matching personal and work accounts — whether by accident or on purpose — becomes much harder.

In practice, this looks like Anthropic’s and Okta’s implementation, which means Claude Managed Agents can now also be imported into the corporate directory and treated as identities with human owners, while a compliance interface feeds risk signals like dormant agents or misconfigured accounts back to security teams.

Identity, not authorization

In today’s announcement, Aaron Parecki, Okta’s director of identity standards, calls the result making the identity provider “a centralized governance plane” for MCP access. That governance plane decides who connects to what. It does not decide whether a given action is allowed, though.

It’s worth noting that the permissions the extension hands over are broad. Whether a particular agent should be allowed to run a specific action on a specific resource at a given moment is a decision that is managed by the policy engines and gateways that now typically sit between an agent and the tools it calls.

What’s next

Beyond Anthropic, Okta and Microsoft, companies like Asana, Atlassian, Canva, Figma, Granola, Linear, and Supabase now support the extension. Slack and a number of other companies will soon add support, too.

Okta is also bringing native support for the protocol to its Auth0 developer platform, letting developers expose their MCP servers without implementing it from scratch.

YOUTUBE.COM/THENEWSTACK

Tech moves fast, don't miss an episode. Subscribe to our YouTube channel to stream all our podcasts, interviews, demos, and more.

Created with Sketch.