LINKSPREED Announces Native MCP Server for UIID v2.13, Bringing Privacy-First Identity to the Age of AI Agents

Albuquerque, New Mexico, October 18, 2026 — LINKSPREED LLC today announced that UIID, its local-first identity protocol, will ship with a native Model Context Protocol (MCP) server as part of UIID v2.13. With this step, UIID becomes one of the first identity layers built to work directly with AI agents. Assistants, copilots and autonomous agents will be able to understand, request and respect a person’s digital identity without ever getting uncontrolled access to it.

More information about UIID is available at uiid.me.

The Shift: AI Is Moving From Answering to Acting

For most of the last few years, AI assistants mainly answered questions. That is changing quickly. Today’s agents book appointments, fill out forms, open accounts, write code, manage files and work across dozens of services for the people they serve. Every one of those actions raises the same question: who is acting, and on whose behalf?

Most of the internet was never designed to answer that question for software agents. Logins were built for humans typing passwords. API keys were built for developers. Neither gives an AI agent a safe, limited and reversible way to act for a real person. LINKSPREED believes identity must become a native part of the agent ecosystem rather than an afterthought, and the UIID MCP server is how the company plans to get there.

What Is the Model Context Protocol?

The Model Context Protocol is an open standard that defines how AI applications connect to external tools, data sources and services. Anthropic introduced it in November 2024 as a universal way to replace fragmented, one-off integrations with a single shared protocol.

Before MCP, every combination of AI model and external system needed its own custom connector. Ten AI applications and ten services could mean up to a hundred separate integrations. This is often called the “N×M problem”. MCP solves it with a shared language: a service builds one MCP server, and any compatible AI application can use it.

The design borrows from the Language Server Protocol, which once did the same for programming languages and code editors. Technically, MCP uses JSON-RPC 2.0 messages and defines three roles:

Hosts: the AI applications that people actually use, such as chat assistants, coding tools or agent platforms.
Clients: the connectors inside a host. Each one maintains a link to a specific server.
Servers: the services that offer capabilities to the AI.

An MCP server can offer three main building blocks:

Tools: actions the AI model can carry out, such as “create an alias” or “verify a credential”.
Resources: context and data that the AI or the user can read.
Prompts: reusable templates and workflows that guide common tasks.

From Experiment to Industry Standard

MCP has grown at remarkable speed. OpenAI and Google DeepMind adopted the protocol, and in December 2025 Anthropic donated MCP to the Agentic AI Foundation, a fund under the Linux Foundation co-founded by Anthropic, Block and OpenAI. That move made MCP a neutral, community-governed standard rather than a single-vendor project.

Usage has followed. The protocol’s maintainers report close to half a billion SDK downloads per month across the main SDKs, with the TypeScript and Python SDKs each passing one billion total downloads. Industry analyses count well over 10,000 public MCP servers, and OpenAI, Google, Microsoft and AWS have all built MCP into their agent stacks.

The latest specification, released on July 28, 2026, is the largest revision in the standard’s history. It introduces a stateless protocol core, so any request can be handled by any server instance, which makes MCP far easier to scale. It also adds header-based routing for gateways, cacheable tool lists, a formal extensions framework, and stronger authorization rules. These include issuer validation and a move away from dynamic client registration toward client metadata documents.

The Security Gap MCP Exposes

MCP’s growth has also revealed a serious weakness. When an AI agent can call tools, read data and act across systems, every connection becomes a possible attack surface. Independent researchers have found exploitable flaws in a large share of public MCP servers. They have documented risks such as tool poisoning, “rug pulls” where a tool’s behavior changes after approval, and cross-server shadowing. US security agencies, including the NSA and CISA, have since issued formal guidance on the topic.

The underlying problem is identity. An agent that acts without a clear, verifiable and limited identity is hard to trust, hard to audit and hard to stop. Handing an agent a person’s main login or a long-lived API key gives it far more power than any single task needs. LINKSPREED designed the UIID MCP server to close exactly this gap.

What Is UIID?

UIID is LINKSPREED’s universal identity protocol. It is built on a simple principle: a person’s identity should belong to them, not to the platforms they use.

UIID creates cryptographically secure identities directly on the user’s device. It does not store passwords or personal data in a central database. Authentication uses on-device biometrics and passkeys, so nothing reusable leaves the device and there is nothing central for attackers to steal. Zero-knowledge proofs let users prove facts about themselves without exposing the underlying data.

At the heart of UIID is a split identity architecture:

The Core ID is the user’s master identity. It stays under their exclusive control.
Aliases are context-specific identities derived from the Core ID. A user can have one alias for a shopping platform, another for a community and another for a work tool.

Because each service only sees its own alias, platforms cannot link a person’s activity across independent services. This stops cross-site tracking at the identity level instead of relying on cookie banners or policy promises.

UIID already has a growing set of components. It offers single sign-on integration for developers, an embeddable login button that adds passwordless sign-in to websites without server-side code, and UIID WRAP. WRAP brings UIID to websites that have not integrated it yet by generating data-minimized aliases and credentials, while the user reviews and approves every submission. Verified human users also get access to Web4 infrastructure, including 100 GB of free storage.

The Announcement: A Native MCP Server in UIID v2.13

With UIID v2.13, LINKSPREED will provide a native MCP server that connects UIID directly to the AI agent ecosystem. Any MCP-compatible host, whether an assistant, a coding tool or an enterprise agent platform, will be able to work with UIID through the same standard interface it already uses for other tools.

The goal is not to give AI agents a person’s identity. The goal is to give them permission to act within clear limits that the person sets, can see and can take back at any time.

Planned capabilities

LINKSPREED plans to design the UIID MCP server around the following principles and functions. Details will be confirmed with the v2.13 release.

Delegated, scoped access instead of full credentials. Agents never receive the Core ID or any reusable secret. Instead, a user grants an agent a narrow, time-limited permission for a defined task, such as signing up for a single service with a fresh alias.
Alias management for agents. When an agent needs to register with a new service, it can request a new alias through MCP. That keeps the user’s real identity and other accounts separate from whatever the agent does.
Human approval for sensitive actions. Important steps, such as creating an identity, sharing a verified attribute or authorizing a login, need explicit confirmation from the user on their own device, secured by biometrics. The agent can prepare the action. Only the person can approve it.
Privacy-preserving verification. Agents can ask questions like “Is this user over 18?” or “Is this a verified human?” and get a cryptographic yes or no. The underlying personal data is never shared.
Clear agent identity. Services working with an agent should be able to recognize that an AI is acting on behalf of a verified UIID user, rather than an anonymous bot. This helps platforms tell legitimate agents apart from abuse.
Transparency and revocation. Users can see which agents hold which permissions and withdraw any of them instantly.
Alignment with the current MCP specification. The server will follow the modern MCP architecture, including the stateless core and the hardened authorization model of the July 2026 specification.

Why This Matters

For users

People increasingly let AI handle everyday digital tasks. Today that often means handing over passwords or giving broad account access to tools they cannot fully control. With UIID’s MCP server, users can delegate work to AI agents and still decide what each agent may do, which identity it uses and how long its access lasts. Their real identity stays on their own device.

For developers

Developers building agents no longer need to create their own identity and permission systems or handle sensitive credentials themselves. One standard MCP connection gives their agent access to privacy-first authentication, alias creation and verification. This lowers both development effort and liability.

For businesses and platforms

Companies are under growing pressure to support AI agents while protecting themselves against fraud, bots and data protection risks. UIID lets them accept agent-driven sign-ups and actions that are tied to a verified human, without collecting more personal data than they need. Data minimization is built into the architecture, which simplifies compliance.

For the broader internet

As AI agents become regular participants online, the internet needs a trust layer that works for both humans and machines. LINKSPREED sees UIID as that layer: one identity system that connects people, their agents and the services they use, based on user control instead of platform control. This is a central part of the company’s Web4 vision.

A Statement From LINKSPREED

“AI agents are becoming the new interface to the internet, but the internet still doesn’t know how to trust them,” said Marc Herdina of LINKSPREED LLC. “Giving an agent your password was never the answer. With the UIID MCP server, people can let AI work for them while their identity stays exactly where it belongs: on their own device and under their own control. MCP gives agents a common language. UIID gives that language a trustworthy identity.”

Availability

The MCP server is planned as part of UIID v2.13. Further technical documentation, integration guides and release details will be published when the release is ready. Developers, platforms and partners who want early access or want to explore an integration are invited to get in touch with LINKSPREED.

Users can create their free UIID Core ID today at uiid.me.

About LINKSPREED

LINKSPREED LLC is a technology company based in Albuquerque, New Mexico, building infrastructure for the next generation of the internet, which it calls Web4. Its products include UIID, a local-first, privacy-first identity protocol that uses on-device biometrics, passkeys and zero-knowledge proofs to replace passwords and centralized identity data. LINKSPREED’s mission is to make people the owners of their digital identities and to provide a trusted base layer for humans, machines and AI agents.

Learn more: uiid.me