MCP server

Connect your AI agent to the gallery.

What is MCP

MCP (Model Context Protocol) is an open standard that lets AI agents call outside tools. With it, your agent can search this gallery itself and work from real pages.

The boringweb server gives it the same library you browse here: every site, every captured page and every section, with the studio credit attached.

What your agent can do

Find sites in an industry
It searches the gallery by industry, page type, size, region, platform or studio.
"Show me law firm sites built on Webflow."
Study one kind of section
It pulls heroes, pricing tables, team grids, footers and other sections from across the gallery.
"Find pricing sections from accounting firms."
Read a whole site
It gets every captured page of a site with its screenshots, sections, fonts and platform.
"What pages does this dental site have, and what fonts does it use?"
Look up a studio
It lists every site in the gallery that a studio designed.
"What else has Field & Fold designed?"

How to install

Pick your tool and add the server with your key. Signed in, the Copy button fills it in for you.

Claude CodeOne command in your terminal.

Run this in your terminal, then start a new Claude Code session.

claude mcp add --transport http boringweb https://mcp.boringweb.gallery/mcp --header "Authorization: Bearer YOUR_KEY"

Claude DesktopAdd a custom connector and paste the server address.

Open Settings, then Connectors. Choose "Add custom connector", name it boringweb and paste this address.

https://mcp.boringweb.gallery/mcp?key=YOUR_KEY

CursorAdd it in Settings under MCP, then restart.

Open Settings, then MCP, and add a new server. Or put this in ~/.cursor/mcp.json and restart Cursor.

{
  "mcpServers": {
    "boringweb": {
      "url": "https://mcp.boringweb.gallery/mcp",
      "headers": { "Authorization": "Bearer YOUR_KEY" }
    }
  }
}

LovablePaste the server address in your project settings.

Open your project settings and find the MCP servers. Add a server, name it boringweb and paste this address.

https://mcp.boringweb.gallery/mcp?key=YOUR_KEY

ChatGPTAdd it as a connector in Settings.

Open Settings, then Connectors. Turn on developer mode, create a connector and paste this address.

https://mcp.boringweb.gallery/mcp?key=YOUR_KEY

CodexOne command in your terminal.

Run this in your terminal, then start a new Codex session.

codex mcp add boringweb --url "https://mcp.boringweb.gallery/mcp?key=YOUR_KEY"

VS CodeWorks with Copilot Chat and any MCP extension.

Run this in your terminal. Or run "MCP: Add Server" from the Command Palette and paste the address.

code --add-mcp '{"name":"boringweb","type":"http","url":"https://mcp.boringweb.gallery/mcp","headers":{"Authorization":"Bearer YOUR_KEY"}}'

ZedAdd it to your context_servers config.

Open your settings file and add the server under context_servers.

"context_servers": {
  "boringweb": { "url": "https://mcp.boringweb.gallery/mcp?key=YOUR_KEY" }
}

Using something else? The server address ishttps://mcp.boringweb.gallery/mcpwith your key sent asAuthorization: Bearer YOUR_KEY

Your keyA key comes with a free account

The tools

Reference lists

list_industries: Industries with their sub-industries and entry counts.
list_page_types: Page types with entry counts.
list_section_types: Section types with section counts.

Search

search_sites: Sites by industry, sub-industry, page type, size, region, platform or studio, by a text query, or by likeness to a site you name.
search_sections: Section crops by section type, industry, sub-industry or page type, or by a text query over what the sections say and show.
search_interfaces: Products from Interfaces by industry, sub-industry, screen type, audience or kind, or by words.

Detail

get_site: The full record for one site, with every captured page and its sections.
get_studio: A studio's name, address and entries in the gallery.
get_interface: One product from Interfaces with its screens, the company that made it and the source.
get_section_spec: How one section is built: its outline, grid and spacing, copy slots, fonts and colours, with desktop and mobile crops.
get_design_md: A site's design tokens and the reasoning behind them, in the DESIGN.md format.

Structure

get_content_model: What a page type holds in an industry: the fields, their usual order and how long each runs.
get_rubric: A short checklist for scoring a draft page against the curated examples.
get_compliance_checklist: The elements usually expected on a site in an industry, with public sources. Not legal advice.
get_stack: The services sites in an industry commonly use, with counts.
plan_site: A starter sitemap for an industry: page types, the sections each page usually has, and three references for each section.

Boards

get_board: A board by its ID, as a brief, with the notes and specs for each pick.

Free text

search: Plain words, read the way the site's search reads them. Returns the filters it matched along with the results.

Pricing

Free, with a rate limit per IP address.

Questions

Stuck setting it up, or want a tool it doesn't have yet?

Get in touch