
We are kicking off 2026 with a set of updates focused on one theme: making authentication easier to adopt and easier to run, whether you are building an MCP server, shipping AI agent workflows, or supporting enterprise SaaS.
This post is a roundup of what’s new across MCP, agent authentication, and our developer experience.
Client onboarding for MCP servers is now simpler. Scalekit supports Client ID Metadata Documents (CIMD), allowing MCP clients to identify themselves via a metadata URL instead of manual pre-registration.
This is designed for ecosystem-facing MCP servers with many independent or distributed clients, where pre-registering every client does not scale. CIMD can be enabled with a single setting in the dashboard.
If you want a deeper dive, watch our CTO Ravi, break down DCR vs. CIMD with Akshay (our in-house expert), who built the feature at Scalekit.
Read more about CIMD on the launch blog.
We are also introducing Agent Auth: a complete solution for outbound authentication when AI agents interact with external systems. This includes agents acting on behalf of users, service accounts, and MCP tools.
Agent Auth includes:
Agent Auth works across major agent frameworks and runtimes, so teams can standardize how outbound auth is handled as their agent architecture evolves.

Learn more about Agent Auth.
You can now provision users programmatically during authentication, with control over user creation, attribute mapping, and organization assignment at auth time.
Get started here.
The new Passkeys SDK lets you build fully custom Manage Passkeys screens while Scalekit handles the underlying security and protocol complexity.
Get started here.
We have restructured both the dashboard and the documentation to better reflect how teams actually build.
You now select your authentication mode upfront: go full-stack with everything integrated, or pick modular components like MCP Auth, Agent Auth, SSO, and SCIM. The documentation mirrors this structure, with use case-driven guides for MCP Auth, Agent Auth, and enterprise authentication, plus an end-to-end guide for using Scalekit as your complete auth layer. Go to Scalekit docs.
.gif)
Getting started is faster with forkable, real-world examples, including an end-to-end Coffeedesk implementation (Go + Node.js) and a separate Django Python implementation.These examples are designed to show realistic authentication flows, not just isolated snippets, so teams can adapt them directly into production codebases. Watch the Coffeedesk demo below. 👇 Or, check out sample apps here.
To make documentation more accessible, we also added Ask AI directly inside the docs. You can write questions in natural language and get clear answers in a few lines, without jumping between pages. Go to Scalekit docs.

These updates are aimed at giving teams clearer choices, fewer bespoke integrations, and a more predictable path from setup to production. If you are building with MCP, experimenting with AI agents, or rolling out enterprise features, we would love to hear what you are working on and where we can make things smoother.