Discovery Endpoints

Every store with public MCP enabled exposes two discovery endpoints for programmatic server configuration: a structured JSON manifest and a plain-text agent guide.

Automatic generation

These endpoints are automatically generated for every store with public MCP enabled. No configuration needed by the merchant. The endpoints reflect the store's current tool visibility settings, commission rates, and active campaigns.

.well-known/mcp.json

A structured JSON manifest describing the store's MCP server capabilities, transport URL, available tools, authentication requirements, and rate limits.

Request

Endpoint

GET https://mcp.chatcast.io/mcp/public/{storePublicId}/.well-known/mcp.json

Property	Value
Authentication	None
Content-Type	`application/json`
Cache-Control	`public, max-age=3600` (1 hour)
Error	Returns 404 if store not found or public MCP is disabled.

Response

{
  "name": "Comet Public Commerce MCP",
  "description": "Public read-only MCP server for AI agents to access product catalog, FAQs, policies, and commission information.",
  "version": "1.0.0",
  "protocol": "mcp",
  "transport": {
    "type": "http+sse",
    "url": "https://mcp.chatcast.io/mcp/public/server/{storePublicId}/mcp"
  },
  "authentication": {
    "type": "none",
    "description": "No authentication required. This is a public, read-only endpoint."
  },
  "capabilities": {
    "tools": true,
    "resources": false,
    "prompts": false
  },
  "tools": [
    {
      "name": "search_products",
      "description": "Search the product catalog using natural language.",
      "category": "discovery"
    },
    {
      "name": "get_product_details",
      "description": "Get complete product details by public ID, with commission metadata.",
      "category": "discovery"
    },
    {
      "name": "get_faqs",
      "description": "Get pre-generated frequently asked questions for a product.",
      "category": "discovery"
    },
    {
      "name": "get_policies",
      "description": "Search store policies using semantic matching.",
      "category": "discovery"
    },
    {
      "name": "get_brand_info",
      "description": "Get brand profile information for the store.",
      "category": "discovery"
    },
    {
      "name": "get_commission_rates",
      "description": "View current commission rates for this store.",
      "category": "commission"
    },
    {
      "name": "get_active_campaigns",
      "description": "Discover currently active commission campaigns.",
      "category": "commission"
    },
    {
      "name": "register_agent",
      "description": "Register your agent for attribution tracking.",
      "category": "commission"
    },
    {
      "name": "get_agent_offers",
      "description": "Get exclusive offers available for registered agents.",
      "category": "commission"
    }
  ],
  "rate_limits": {
    "description": "100 requests per minute per IP address",
    "headers": [
      "X-RateLimit-Limit",
      "X-RateLimit-Remaining",
      "X-RateLimit-Reset"
    ]
  }
}

Response Fields

Field	Type	Description
`name`	string	Server name identifier.
`version`	string	MCP server version (currently 1.0.0).
`transport.type`	string	Transport protocol. Always `http+sse` (Streamable HTTP).
`transport.url`	string	The MCP endpoint URL to connect to.
`authentication.type`	string	Always `none` for the public endpoint.
`capabilities`	object	MCP capabilities. Tools are enabled; resources and prompts are not.
`tools`	array	List of available tools with name, description, and category.
`rate_limits`	object	Rate limiting details and HTTP headers to inspect.

Use Case

Use .well-known/mcp.json for programmatic auto-configuration. IDE plugins, MCP client registries (like Smithery), and automated agent frameworks can fetch this endpoint to discover the server URL, available tools, and connection requirements without manual configuration.

llm.txt

A plain-text agent guide designed for AI agents to quickly understand a store's MCP capabilities, available tools, commission rates, and active campaigns.

Request

Endpoint

GET https://mcp.chatcast.io/mcp/public/{storePublicId}/llm.txt

Property	Value
Authentication	None
Content-Type	`text/plain; charset=utf-8`
Cache-Control	`public, max-age=3600` (1 hour)
Error	Returns 404 if store not found or public MCP is disabled.

Response

Response (plain text)

# Cool Coffee Co — AI Agent Endpoint

MCP Server URL: https://mcp.chatcast.io/mcp/public/server/demo/mcp
Discovery JSON: https://mcp.chatcast.io/mcp/public/demo/.well-known/mcp.json
Protocol: Model Context Protocol (MCP)
Transport: HTTP+SSE (streamable)
Authentication: None required (public, read-only)

## Available Tools
  - search_products: Search the product catalog using natural language.
  - get_product_details: Get complete product details by public ID, with commission metadata.
  - get_faqs: Get pre-generated frequently asked questions for a product.
  - get_policies: Search store policies using semantic matching.
  - get_brand_info: Get brand profile information for the store.
  - get_commission_rates: View current commission rates for this store.
  - get_active_campaigns: Discover currently active commission campaigns.
  - register_agent: Register your agent for attribution tracking.
  - get_agent_offers: Get exclusive offers available for registered agents.

## Agent Commission Program
Default commission rate: 8.0%

## Active Campaigns
  - Spring Sale: +5.0% bonus, 10.0% customer discount (code: SPRING10)

## Getting Started
1. Connect to the MCP server URL above using any MCP-compatible client.
2. Call register_agent to receive attribution tracking.
3. Use search_products, get_faqs, and get_policies to help users.
4. Earn commissions on referred purchases.

Powered by ChatCast — https://mcp.chatcast.io

Content Sections

The llm.txt file is dynamically generated and includes the following sections:

Section	Content
Header	Store name, MCP server URL, discovery URL, protocol, and transport details.
Available Tools	List of all publicly visible tools with brief descriptions.
Agent Commission Program	Default commission rate percentage (if configured).
Active Campaigns	Currently active campaigns with bonus rates and customer discounts. Only shown if campaigns exist.
Getting Started	Quick-start instructions for connecting and using the server.

Use Case

Use llm.txt when you want a quick, token-efficient overview of a store's MCP capabilities. AI agents that prefer plain text over JSON can consume this endpoint to understand the store's offerings, commission structure, and how to get started -- all in a single request.

When to Use Each Endpoint

Endpoint	Best For	Format
`.well-known/mcp.json`	Programmatic auto-configuration: IDE plugins, MCP registries, automated agent frameworks.	JSON
`llm.txt`	AI agents that want a quick plain-text overview of the store's capabilities, tools, and commission rates.	Plain text

Tool Visibility

Dynamic tool list

Merchants can toggle individual tools on or off via the brand console. Both discovery endpoints reflect the current visibility settings. A tool listed in the discovery response may later be hidden (and vice versa). Always check the discovery endpoint for the most current list of available tools.

Example: Fetching Discovery

You can fetch either endpoint with a simple HTTP request:

curl

# Fetch the structured JSON manifest
curl https://mcp.chatcast.io/mcp/public/demo/.well-known/mcp.json

# Fetch the plain-text agent guide
curl https://mcp.chatcast.io/mcp/public/demo/llm.txt

Replace demo with the actual store public ID.