Implement Impresee Smart Search MCP in your e-commerce
Connect AI agents to your Impresee’s search intelligence through a simple implementation, technical documentation, and ready-to-use examples.
What is Impresee Smart Search MCP?
Impresee Smart Search MCP allows AI agents to query your smart search engine to answer product-related questions using your e-commerce site’s search logic.
Each enabled store gets a stateless MCP server reachable at:
https://api.impresee.com/MCP/search/{store_public_id}The store_public_id identifies the store. The server is stateless — each request is independent, and no session is kept between calls.
What can you build?
Shopping Assistant
Customers can find products through conversations.
Chatbots
Intelligent catalog-based searches and answers.
Copilots
Sales Team Assistants.
Internal Agents
Inquiries regarding support or inventory.
Installation Guide
URL extension
What is the URL extension?
The URL preview below the field updates live as you type so you can verify how it will look before saving.
A short identifier appended to the MCP endpoint URL to identify the store. It must contain only letters, digits, and underscores ([a-zA-Z0-9_]+). A live URL preview updates below the field as you type.
Active MCP URL
When the saved configuration is enabled, a green banner appears on Dashboard Pro with the live endpoint URL. This URL is read-only — it reflects the last saved state, not unsaved edits. Clients (AI assistants, MCP host apps) connect to this URL to access the store’s tools and prompts.
https://api.impresee.com/MCP/search/impresee
Tools
Impresee Smart Search MCP offers a variety of tools. You can enable or disable them to grant access to different features.
Searches Tool
Let the AI run product search queries against your store. The AI sends:
{
"query_keywords": "red dress",
"category": "women"
}query_keywords is required. category is optional — it restricts results to a main category when provided. The response includes product results with rank, relevance, and catalog attributes.Categories Tool
Returns the store’s main categories so the AI can navigate the product taxonomy before searching. The AI calls it with no input and receives:
{
"categories": ["women", "men", "shoes"]
}Useful when the shopper’s request is broad — the AI can first orient itself, then search within the right category.
Prompts
What are prompts?
Prompts are pre-built conversation templates that AI clients can invoke by name. They define the instructions the AI receives and any dynamic arguments the caller must supply at runtime.
Disabled prompts are not served to AI clients but remain in the configuration so you can re-enable them later. Deleted prompts are permanently removed from the database on the next save.
Arguments JSON format
An array of objects, each defining one dynamic input:
[
{
"name": "user_request",
"title": "User request",
"description": "What the shopper is looking for",
"required": true
}
]name is required and must be a simple identifier (letters, digits, underscores). required must be a boolean — true or false, not a string.Texts JSON & placeholders
role ("user" or "assistant") and a text:[
{
"role": "user",
"text": "Find me ${user_request}"
}
]${argument_name} or {{argument_name}} to embed argument values. The preview below the field highlights placeholders in orange. Warnings appear if a placeholder has no matching argument or if a required argument is never used.Save
What happens when I save?
1. Local validation — URL extension format and all prompt JSON fields are checked. Errors are shown inline, and saving is blocked until fixed.
2. Server validation — the config is sent to the backend. If an error is returned, it is shown, and saving is aborted.
3. Upload — the config is written to the database. The backend then signals all active MCP servers to reload their in-memory state so changes take effect immediately.
