LLMs & AI Agents

LLMs & AI Agents

Little Bird Electronics is an AI-first store. Every product page is machine-readable, every API is open, and AI agents can shop our store as easily as a human. Whether you're using Claude, ChatGPT, or a custom agent, here's how it all works.

llms.txt — the starting point

Our /llms.txt file is a machine-readable index of the entire store — products, collections, policies, and API documentation. LLMs and AI agents use this as the entry point to understand what's available.

GET /llms.txt

This file follows the llms.txt specification and provides a structured overview that fits within an LLM's context window.

How AI agents shop our store

1. Discover — The agent fetches /llms.txt to understand the store, or searches /products.md?q= for specific items.
2. Research — Get detailed product info including variants, real-time stock levels, pricing, and specifications via /products/{handle}.md.
3. Compare — Use the JSON API to compare products, check collection listings, or filter by availability.
4. Add to cart — Using WebMCP tools or the cart API, the agent adds variants to the cart.
5. Checkout — The agent can get shipping estimates and present the cart summary. The human completes payment.

WebMCP tools

Our store registers WebMCP tools via navigator.modelContext on every page. Browser-based AI agents (Chrome 146+) can call these tools to interact with our store programmatically.

Available tools:

• search_products — Search catalog by keyword
• get_product_details — Full product info by handle
• add_to_cart — Add a variant to the cart
• get_cart — View current cart contents
• update_cart_item — Change quantity or remove
• remove_from_cart — Delete item from cart
• list_collections — List all categories
• get_collection_products — Products in a collection
• get_shipping_estimate — Shipping rates by postcode
• get_store_info — Contact, policies, payment info

When an AI agent visits any page on our store in a WebMCP-compatible browser, it automatically discovers these tools. No scraping required — the agent can search, check availability, add items to cart, and get shipping estimates through structured tool calls.

JSON API endpoints

Every product and collection page responds to .json format for programmatic access:

# Search products
GET /products.json?q=coral+tpu

# Product details with variants and stock
GET /products/{handle}.json

# Collections
GET /collections.json
GET /collections/{handle}.json

# Cart
GET /cart.json
POST /cart/add variant_id=123&quantity=1

For the full read-only API, see Public API.

Structured data on every page

Every product page includes rich JSON-LD structured data (Schema.org Product, Offer, AggregateRating, Review, BreadcrumbList). AI agents, Google Shopping, and price comparison tools can all parse this data directly from the HTML.

Downloadable formats

Products, quotes, invoices, and the cart are all available as Markdown, CSV, and PDF:

• /products/{handle}.md — Product page as Markdown
• /products.md?q={search} — Product catalog as Markdown
• /account/quotes/{id}/export_md — Quote as Markdown
• /account/orders/{id}/invoice_md — Invoice as Markdown
• /cart/export_md — Cart as Markdown

These formats are optimised for LLM context windows — clean, structured, and concise.