KnitKnot
You are reading the agent-optimized layer of this page: the literal markdown we serve to AI crawlers and assistants, shipped in the page source of every visit. Making sure AI reads the right facts about a company is literally what KnitKnot does.

# What is WebMCP?

WebMCP is an in-browser adapter that lets a compatible agent discover read-only tools exposed by the page. KnitKnot keeps those tools tied to the same fact ledger.

## What is WebMCP?

WebMCP is an in-browser adapter for compatible agents. A small page script registers read-only tools such as listing approved resources, reading one resource, and describing the active agent interfaces.

The tools return views of the same approved fact ledger used by the page and its other agent interfaces. The adapter is not a new authoring system and does not give an agent permission to alter the customer's content.

## How is it different from `/mcp`?

The customer-domain `/mcp` surface is a read-only HTTP interface for agents that can call the domain directly. WebMCP registers a small set of read tools inside a compatible browser session. Both are interface doors over the same facts.

KnitKnot also operates an authenticated workspace MCP server for customer assistants. That server exposes authorized benchmark and workflow data. It is separate from the public, read-only interface on a managed customer domain.

## What does WebMCP promise?

It provides a governed way for a compatible in-browser agent to request approved information. It does not promise that a consumer agent supports the adapter, chooses a tool, cites a page, or changes its answer.

## Is it generally available?

The WebMCP adapter is built and rolling out behind deployment gates. KnitKnot verifies the customer domain, eligible pages, and enabled interfaces before launch.

Next: [agent-ready content](/learn/what-is-agent-ready-content/) or [agent interfaces](/product/agent-interfaces/).

Raw mirror of this content: https://knitknot.ai/learn/webmcp.md. Site-wide summary: /llms.txt · full content: /llms-full.txt

← Guides
Published
Reading time
2 minutes

Guide

What is WebMCP?

WebMCP is an in-browser adapter that lets a compatible agent discover read-only tools exposed by the page. KnitKnot keeps those tools tied to the same fact ledger.

A source trail linking claim fragments to citations and evidence receipts

What is WebMCP?

WebMCP is an in-browser adapter for compatible agents. A small page script registers read-only tools such as listing approved resources, reading one resource, and describing the active agent interfaces.

The tools return views of the same approved fact ledger used by the page and its other agent interfaces. The adapter is not a new authoring system and does not give an agent permission to alter the customer’s content.

How is it different from /mcp?

The customer-domain /mcp surface is a read-only HTTP interface for agents that can call the domain directly. WebMCP registers a small set of read tools inside a compatible browser session. Both are interface doors over the same facts.

KnitKnot also operates an authenticated workspace MCP server for customer assistants. That server exposes authorized benchmark and workflow data. It is separate from the public, read-only interface on a managed customer domain.

What does WebMCP promise?

It provides a governed way for a compatible in-browser agent to request approved information. It does not promise that a consumer agent supports the adapter, chooses a tool, cites a page, or changes its answer.

Is it generally available?

The WebMCP adapter is built and rolling out behind deployment gates. KnitKnot verifies the customer domain, eligible pages, and enabled interfaces before launch.

Next: agent-ready content or agent interfaces.