> Markdown version of [/videos/1811-webmcp-making-agents-a-first-class-citizen-of-the-web-andre-cipriani-bandarra-francois-beaufort?t=519](https://www.wearedevelopers.com/videos/1811-webmcp-making-agents-a-first-class-citizen-of-the-web-andre-cipriani-bandarra-francois-beaufort?t=519). Every page supports `.md` or `Accept: text/markdown`. Links point to the HTML versions so they work for humans too. Agent guide: [/agents.md](https://www.wearedevelopers.com/agents.md).

---

# WebMCP - Making Agents a First-Class Citizen of the Web - Andre Cipriani Bandarra & François Beaufort

Brittle DOM scraping is dead. Andre and François reveal how the new WebMCP standard lets developers expose native functions, making AI agents first-class citizens of the web.

- **Speakers:** Andre Cipriani Bandarra, François Beaufort
- **Event:** Coffee With Developers
- **Published:** February 16, 2026
- **Duration:** 30:42
- **URL:** https://www.wearedevelopers.com/videos/1811-webmcp-making-agents-a-first-class-citizen-of-the-web-andre-cipriani-bandarra-francois-beaufort

## Summary

The modern web was expressly designed for human psychology and user interfaces, forcing AI agents to effectively "hack" their way through sites using automated browsers and complex DOM scraping. This approach is highly prone to errors, wastes expensive LLM context tokens, and often resembles bot-like, brute-force behavior that website owners want to block. To bridge this gap, WebMCP (Web Model Context Protocol)—a proposed W3C standard backed by organizations including Anthropic, Google, and Microsoft—enables websites to expose functions natively to browser-based agents, elevating AI to a first-class citizen of the web.

WebMCP introduces both an imperative JavaScript API and a lightweight declarative approach that annotates standard HTML forms with `tool-name` and `tool-description` attributes. By adopting these semantic annotations, developers can guide AI agents precisely instead of relying on them to guess form intent or reverse-engineer layout elements. Currently available for testing in Chrome Canary behind an experimental flag, the protocol enables agents to instantly discover available tools, understand expected parameters, and flawlessly execute actions like booking reservations or fetching dynamic content.

To facilitate adoption and debugging, a Chrome extension called the Model Context Tool Inspector allows developers to manually test their exposed functions and observe how a basic local LLM interprets the registered tools. The introduction of WebMCP creates a mutually beneficial ecosystem: website owners retain structured control over programmatic access (protecting their sites from aggressive, unstructured scraping), while agents execute tasks faster and can dynamically autocorrect user input errors based on precise validation feedback. Ultimately, building a standard, semantically structured page stands to become the new SEO for the agentic web.

**Keywords:** WebMCP standard, ai browser agents, programmatic web access, dom scraping alternatives, w3c community protocols, declarative html forms, imperative javascript apis, chrome canary testing, model context tool inspector, token usage optimization, llm function calling, agentic web seo, semantic web standards, intelligent form validation, ai agent tool registration

## Chapters

1. **The problem with agents acting like humans on the web** (00:01) — Allowing bots to scrape and interact with web interfaces built for human psychology creates unnecessary overhead and destructive behavior.
1. **Introduction to the model context protocol and WebMCP** (01:41) — This standard allows browser-based agents to interact directly with website tools through declarative or imperative coding patterns.
1. **Testing the imperative API in Chrome with an extension** (04:31) — Enabling the testing flag allows developers to expose and interact with programmable JavaScript tools using a browser extension.
1. **Annotating web forms using the WebMCP declarative API** (06:26) — Adding tool name and description attributes to HTML form elements enables seamless programmatic agent interactions without relying on custom JavaScript.
1. **Debugging WebMCP implementations using the tool inspector extension** (08:39) — Developers can verify exposed functions, invoke them manually, and test prompt engineering scenarios using a lightweight local model.
1. **Reducing server load and token consumption from agent scraping** (10:26) — Providing direct programmatic access prevents destructive scraping behaviors while optimizing large language model token usage for automated workflows.
1. **Gathering developer feedback for the early WebMCP preview** (11:53) — API designers seek insights on edge cases, tool registration patterns, and prompt engineering expectations to refine the specification.
1. **Optimizing content access without blocking automated agent traffic** (14:22) — The protocol serves as a focused path for willing agents to parse functionality rather than a security mechanism for blocking undesired crawling.
1. **Accessing the Chrome extension and open source code repository** (17:22) — Developers can install the model testing module from the web store or contribute directly via the open source repository to run experiments.
1. **Protecting browser platform health during the rise of agents** (20:27) — As automated networks increasingly navigate the internet, maintaining core platform integrity requires balancing performance optimizations with traditional human browsing.
1. **Driving better web semantics and AI prompt auto-correction** (24:30) — Returning structured validation errors to an automated system allows it to auto-correct inputs without human intervention while encouraging stronger semantic markup.
1. **Reviewing the W3C community group early draft proposal** (26:52) — Web engineers are encouraged to review the public W3C draft specification and join the official experimental mailing list to shape the final standard.

## Related Moments

- [Standardizing agent interactions with the Web MCP proposal](https://www.wearedevelopers.com/videos/1805-wearedevelopers-live-modern-devops-for-iot-devices-and-more) (from "WeAreDevelopers LIVE - Modern DevOps for IoT Devices and More")
- [Automating tasks securely through agentic browser protocols](https://www.wearedevelopers.com/videos/100014-what-s-new-in-web-2026-edition) (from "What’s New in Web? 2026 Edition")
- [Transforming interactive website elements into automated WebMCP tools](https://www.wearedevelopers.com/videos/100199-making-documentation-ai-ready-preparing-your-docs-for-the-llm-era) (from "Making Documentation AI-Ready: Preparing Your Docs for the LLM Era")
- [Introduction to DevTools MCP and features](https://www.wearedevelopers.com/videos/100206-prompt-render-debug-ai-powered-react-development-with-chrome-devtools) (from "Prompt, Render, Debug: AI-Powered React Development with Chrome DevTools")
- [Core concepts and advantages of the Model Context Protocol](https://www.wearedevelopers.com/videos/100305-api-mcp-or-mcp-app-choosing-the-right-surface-for-ai-agents) (from "API, MCP or MCP App? Choosing the right surface for AI agents")
- [Standardizing agent tool discovery with MCP](https://www.wearedevelopers.com/videos/100336-from-apis-to-mcp-enterprise-governance-registry-and-controls) (from "From APIs to MCP: Enterprise Governance, Registry, and Controls")

## Related Articles

- [WebMCP: Empowering Agents as First-Class Citizens of the Web](https://www.wearedevelopers.com/magazine/696-webmcp-empowering-agents-as-first-class-citizens-of-the-web)
- [The Web We Broke (And Why AI Agents Are Paying the Price) - AgentCon Berlin](https://www.wearedevelopers.com/magazine/735-the-web-we-broke-and-why-ai-agents-are-paying-the-price-agentcon-berlin)
- [Everything a Developer Needs to Know About MCP with Neo4j](https://www.wearedevelopers.com/magazine/604-everything-a-developer-needs-to-know-about-mcp-with-neo4j)
- [Dev Digest 212: WebMCP or MCP, What is DevRel, AI's 10% Productivity Boost, and a 49MB Web Page…](https://www.wearedevelopers.com/magazine/717-dev-digest-212-webmcp-or-mcp-what-is-devrel-ai-s-10-productivity-boost-and-a-49mb-web-page)

## Related Jobs

- [Senior Backend Developer — AI: MCP & Agent Engine](https://www.wearedevelopers.com/jobs/48297-senior-backend-developer-ai-mcp-agent-engine) at **basebox GmbH**
- [Principal Product Manager, Agent Platform](https://www.wearedevelopers.com/jobs/ext/277541-principal-product-manager-agent-platform) at **GitHub**
- [AI Software Engineer (Germany)](https://www.wearedevelopers.com/jobs/48317-ai-software-engineer-germany) at **Sunhat**
- [Senior AI Agent Software Engineer (Go, Python) (m/f/x)](https://www.wearedevelopers.com/jobs/48277-senior-ai-agent-software-engineer-go-python-m-f-x) at **Dynatrace**
- [Mid/Senior Full-Stack Engineer (Web-first)](https://www.wearedevelopers.com/jobs/ext/1210833-mid-senior-full-stack-engineer-web-first) at **SMG Swiss Marketplace Group**
- [Senior Web Designer, Growth](https://www.wearedevelopers.com/jobs/ext/102722-senior-web-designer-growth) at **Intercom, Inc.**