Use BlindsBazaar in Claude, Codex & Any AI Assistant

BlindsBazaar ships a Model Context Protocol (MCP) server: a small, read-only bridge that lets AI assistants answer real questions from our live store — exact made-to-measure prices for your window size, product search, our FAQs and delivery terms. No secrets, no accounts, no ability to order or change anything.

Reviewed November 2026.

What your AI can do with it

Eleven tools, all backed by the same public storefront API the website uses — so nothing the assistant says can drift from what the site actually shows or charges.

search_products
Search the live catalog by text, blind type, room, collection, opacity or price cap.
get_product
Full product detail: prices, size ranges, materials, photos, colourways, ratings.
quote_blind
Price an exact window size — the store's own pricing engine, the same one the checkout charges.
list_blind_types
Blind types with the rate card: ₹/sq ft, minimum charge, motorizable.
list_rooms / list_collections
Taxonomy slugs for filtering, with storefront URLs.
get_configurator_options
Minimum billable area plus fabric, control and mount options with ₹ add-ons.
search_faqs
The store's published FAQs — measuring, pricing, delivery, warranty.
search_journal / get_journal_story
The BlindsBazaar Journal: buying guides, care guides, city guides.
delivery_coverage
Where orders are delivered, making and transit times, self-install policy.

Step 1 — nothing to install

BlindsBazaar runs its MCP server for you. Add our hosted endpoint as a remote MCP server and your AI client is connected — no cloning, no build steps, no code on your machine:

https://api.blindsbazaar.in/mcp

The endpoint is read-only: it can search the catalog, quote exact window prices, and look up FAQs and delivery info — but it cannot place orders, edit anything, or access accounts. Every client below shows how to add it.

Prefer running it yourself? Install it as a local stdio server with pip install blindsbazaar-mcp or uvx blindsbazaar-mcp (Python 3.11+, published on PyPI), and use the "command"-style snippets shown for some clients.

Step 2 — add it to your AI client

Most clients connect with a single URL. Pick your client below — the hosted endpoint needs no environment variables and no paths.

Claude Desktop

Settings → Connectors → Browse connectors → Add custom connector

Add a custom connector with this URL (nothing to install):

https://api.blindsbazaar.in/mcp

Prefer a local process? Edit the config file instead (Windows: %APPDATA%Claudeclaude_desktop_config.json; macOS: ~/Library/Application Support/Claude/claude_desktop_config.json):

{
  "mcpServers": {
    "blindsbazaar": {
      "command": "uvx",
      "args": ["blindsbazaar-mcp"]
    }
  }
}

Claude Code

Any project, from the terminal

One command registers the hosted server:

claude mcp add --transport http blindsbazaar https://api.blindsbazaar.in/mcp

Add --scope user to make it available in every project.

Verify with claude mcp list, then just ask about blinds in the session.

OpenAI Codex CLI

~/.codex/config.toml

Point Codex at the hosted server (restart Codex after editing):

[mcp_servers.blindsbazaar]
url = "https://api.blindsbazaar.in/mcp"

Codex discovers the tools automatically; ask it to quote a window or look up a fabric.

Cursor

File → Cursor Settings → MCP & Integrations → Add custom MCP server

Paste the hosted URL:

https://api.blindsbazaar.in/mcp

Cursor detects the tools immediately; enable the server's toggle in the MCP list.

Windsurf

Windsurf Settings → MCP → Add server

Add the hosted URL as a streamable HTTP server:

https://api.blindsbazaar.in/mcp

Click Refresh in the MCP panel after saving.

VS Code (GitHub Copilot)

Command Palette → MCP: Add Server…, or commit .vscode/mcp.json with the repo

VS Code uses a servers object with an http type:

{
  "servers": {
    "blindsbazaar": {
      "type": "http",
      "url": "https://api.blindsbazaar.in/mcp"
    }
  }
}

Copilot Chat picks the tools up in agent mode — no restart needed.

ChatGPT

Settings → Connectors → Create

Create a custom connector with this URL and enable it in the conversation:

https://api.blindsbazaar.in/mcp

Cline (VS Code extension)

Cline panel → MCP Servers → Configure

Standard mcpServers JSON with the remote URL:

{
  "mcpServers": {
    "blindsbazaar": {
      "url": "https://api.blindsbazaar.in/mcp"
    }
  }
}

Toggle the server on in Cline's MCP list after saving.

Gemini CLI

~/.gemini/settings.json

mcpServers object with the remote URL:

{
  "mcpServers": {
    "blindsbazaar": {
      "url": "https://api.blindsbazaar.in/mcp"
    }
  }
}

pi / DeepAI Code

In any chat, just ask

The pi harness discovers MCP tools automatically. Or add it explicitly in the MCP settings with this URL:

https://api.blindsbazaar.in/mcp

Run a local copy against a local backend (development)

Working on the store itself? Run the local stdio server pointed at your local API instead of production:

{
  "mcpServers": {
    "blindsbazaar": {
      "command": "uvx",
      "args": ["blindsbazaar-mcp"],
      "env": { "BLINDSBAZAAR_API_URL": "http://localhost:8001/api/v1" }
    }
  }
}

Step 3 — ask it things

  • What would a 6 ft 3 in × 5 ft 6 in blackout roller blind for a bedroom cost?
  • Which zebra blinds suit a living room, and what do they cost per square foot?
  • How do I measure a window for an outside-mount blind?
  • Do you deliver to Chennai, and how long does an order take?
  • Compare our two blind types and their rate cards.

Questions about the MCP server

What is the BlindsBazaar MCP server?

It is a small, read-only server that speaks the Model Context Protocol (MCP). AI assistants like Claude, Codex and Cursor use it to look up our live catalog, quote exact window prices, and answer questions from our FAQs and guides — always from the same data the website itself uses.

Is it safe? Can an AI place an order or change anything?

No. The server is read-only: every tool calls public, unauthenticated endpoints and none of them can place orders, edit products or access accounts. It stores no credentials and exposes no secrets.

Where do the prices come from?

From the store's own pricing engine — the same endpoint the checkout uses. A quote for your exact window size (width and height in feet and inches) matches what you would be charged, including GST, the minimum billing area and the free fitting kit and pelmet.

Does it work with any AI assistant?

Any client that supports MCP: Claude Desktop, Claude Code, OpenAI Codex CLI, Cursor, Windsurf, VS Code with GitHub Copilot, ChatGPT connectors, Cline and Gemini CLI, among others. The setup is the same everywhere: add our hosted URL (https://api.blindsbazaar.in/mcp) as a remote MCP server.

Can I run it locally instead of using the hosted URL?

Yes. Install it as a local stdio server with pip install blindsbazaar-mcp (or uvx blindsbazaar-mcp) and register it with a command-style config. Developers can also point it at a local API by setting the BLINDSBAZAAR_API_URL environment variable.

What questions can I ask once it is connected?

Things like: what would a 6 ft 3 in × 5 ft 6 in blackout roller blind for a bedroom cost; which zebra blinds suit a living room and what do they cost per square foot; how do I measure a window for an outside-mount blind; do you deliver to Chennai and how long does it take.

Good to know

  • The server is read-only by design: there is deliberately no tool that can place orders, touch accounts or edit products, so an AI assistant connected to it cannot spend money.
  • Pricing is never calculated in the AI itself — quotes come from the store's own engine, so the site, the checkout and your assistant always agree.
  • Made-to-measure blinds are priced per square foot with GST included, billed on a minimum area, and ship with a free installation kit and pelmet — see our price guide and measuring guide.