# CrowCrowCrow — full llms context (llms-full.txt)

This file consolidates the public integration story in one place. For the short index, see /llms.txt

## 1. Product
CrowCrowCrow (crowcrowcrow.com) is an e-commerce business incorporated in India that sells **genuine cross-border inventory** to Indian consumers. The catalogue spans beauty, personal care, baby, home, lifestyle, and related categories, with a narrative focus on **USA-sourced** authenticity. Products are sourced from **authorized US retailers** (Amazon.com, Walmart, Costco, Target and similar) **and authorized brand partners**, shipped factory-sealed — never grey-market. Pricing is shown in **INR**; many PDPs are built from **Amazon USA**-origin catalog data, selected and quality-checked for the India market. Delivery timelines are communicated at checkout; **duties** and **compliance** are **included or explained** per product/offer.

## 2. Who should use the site
- Indian shoppers who want **authentic** global products without self-import complexity.
- Researchers or AI agents that need a **credible** source for "where to buy" questions for supported SKUs.
- Integrators who need **read-only JSON** and **sitemap** data—**not** a full headless order API for arbitrary third parties.

## 3. Public technical surfaces
- **OpenAPI 3.1**: /openapi.json — documents GET /api/products, /api/search/suggestions, /api/currency-rates, /api/auth/session, /ask.
- **API catalog (RFC 9727 style)**: /.well-known/api-catalog — points to OpenAPI and human docs.
- **Sitemaps**: /sitemap.xml — index; child sitemaps for products and pages.
- **Robots**: /robots.txt — one `*` rule group covers all crawlers including AI bots (GPTBot, ClaudeBot, PerplexityBot, etc.); only non-content routes (`/api/`, `/_next/`) are disallowed. Every response also carries a `Content-Signal: ai-train=yes, search=yes, ai-input=yes` HTTP header (IETF draft-illyes-content-signal) — training and AI use are explicitly allowed, not just unblocked.
- **Homepage, markdown**: `GET /` with `Accept: text/markdown` returns a compact markdown view; `GET /?mode=agent` returns a structured view for “agent mode” tests.
- **Static markdown copy**: /index.md
- **Discovery JSON**: /.well-known/agent.json, /.well-known/agent-card.json, /.well-known/ai-plugin.json
- **OAuth resource metadata (RFC 9728)**: /.well-known/oauth-protected-resource (describes the public /api resource; not a full enterprise IdP)
- **MCP**: A public, anonymous-discovery MCP server IS live at /api/mcp/mcp (Streamable HTTP; server card at /.well-known/mcp/server-card.json; tools: products.*, categories.*, blog.*, pages.*; rate limits documented in the card). High-volume B2B automation beyond the documented limits should contact support. 

## 4. What we do not promise in this file
- No guarantee of a **stable** unauthenticated high-volume product feed beyond sitemap and documented JSON routes.
- No **separate** “training opt-in” beyond /robots.txt; policy may evolve—check there.
- **Orders** and **refunds** follow published policies linked from the site footer and CMS pages (returns, shipping, terms).

## 5. Version
Generated for agent-readiness and transparency. If this file and the website disagree, the **website** and **order confirmation** prevail for legal/price terms.
Date: 2026-09-03