Human approval
Changes are reviewed before they are applied.
OPEN SOURCE INFRASTRUCTURE FOR THE AGENTIC WEB
WebMCPify helps new and existing web applications become agent-ready by discovering their capabilities, generating WebMCP tools, testing real agents, and independently verifying that actions actually work.
Open sourceNew or existing appsHuman approvedIndependently verified
01 / THE SHIFT
Most websites were designed around pages, buttons, forms, and visual interfaces. AI agents can interact with them, but they often have to infer what the application can do from its UI, DOM, and other signals.
That makes execution slower · more expensive · fragile · difficult to verify
02 / WHY WEBMCPIFY
WebMCP gives new and existing web applications a way to expose structured capabilities to AI agents. A new application can build agent capabilities into its workflow from the start. An existing application can discover, add, and verify them without rebuilding the product.
WebMCPify handles the process around that transition.
03 / PRODUCT WORKFLOW
04 / THE DIFFERENTIATOR
A generated WebMCP tool can look correct in source code and still fail when an agent actually uses it. WebMCPify separates agent execution from independent verification.
05 / TRUST & CONTROL
Built for developers and businesses that need agents to act reliably—not just generate code.
Changes are reviewed before they are applied.
Generation and repair can happen without directly modifying the target application.
Only approved patches are applied.
Success is determined from the application's actual state.
06 / A NEW SECURITY PROBLEM
When websites become executable by AI agents, new risks emerge around prompt injection, tool poisoning, unintended actions, data leakage, authorization boundaries, and accountability.
WebMCP can expose capabilities. Businesses still need to control how those capabilities are used.
WebMCPify is designed around approval, controlled execution, verification, and evidence.
07 / WHO IT'S FOR
WebMCPify works with a new product being designed for agent interaction and with an existing application—the product, workflows, and domain knowledge you have already built.
Let agents operate real product workflows instead of just reading product pages.
↗Enable agents to discover products, manage carts, and complete workflows.
↗Make booking and service capabilities accessible to agents.
↗Make new or existing applications agent-ready without rebuilding the product experience.
↗08 / REAL PRODUCT
Explore the working demonstration: a web application, discovery, generated capabilities, approval, agent execution, and independent verification. The same workflow can start with a new build or an existing site.
Discover products, add to cart, and verify state.
09 / OPEN SOURCE
WebMCP is an emerging technology. The fastest way to build an ecosystem around it is to make the tooling accessible to developers first.
WebMCPify is distributed as open source and available through npm. The goal is to make agent-readiness accessible before the ecosystem becomes mainstream.
Import a web app from GitHub or GitLab, run the WebMCPify workflow in a managed workspace, and review the evidence when it's ready.
10 / ORIGIN
“How can an AI agent reliably use an existing website?”
That question led to building a workflow around the transition:
Built by engineers with years of experience building web products, infrastructure, and agent systems across JavaScript/TypeScript, Python, and emerging AI technologies.
11 / START HERE
The agentic web is emerging. Your application shouldn't have to be rebuilt to participate in it.