Skip to content
vviinn DocsDocs Beta

Model Context Protocol (MCP) Tools Overview

Model Context Protocol (MCP) tools provide specialized capabilities for AI agents to interact with the vviinn platform. These tools enable agents to perform various tasks, from retrieving detailed product information to discovering catalog structures and managing conversational memory.

This is a developer-facing surface: there is no Panel screen for enabling or configuring MCP tools. They use the same API credentials your site already has under Site → API Credentials in Panel — a read-only token for search/discovery tools, and the site’s Basic authentication username/password for tools that require it (memory tools). See API credentials for where to find and rotate those credentials.

Tools by Category

Section titled “Tools by Category”

Search & Discovery

Section titled “Search & Discovery”

These tools enable agents to initiate product searches and retrieve recommendations. See Search & Discovery for the request fields they share.

  • 🔍 text_search

    Embedding-based product retrieval from free-text queries.

  • 📷 visual_search

    Searches based on images, leveraging object detection, OCR, and barcode reading.

  • ✨ similar_products

    Product recommendations based on similarity to a reference product.

  • 🤝 cross_selling

    Complementary product recommendations for a reference product.

Schema

Section titled “Schema”

These tools reveal the structure of a merchant’s catalog — which columns it exposes and what values they hold — so an agent can filter, build facets, and validate shopper input before issuing a search.

  • 🧭 Relevant Columns

    List which product columns are relevant for filtering and display on a given site.

  • 🔬 Describe Column

    Report the catalog-wide values (enumerated options, numeric ranges, category paths) for a single column.

  • 🔐 Public Columns

    The site-wide baseline of which columns can surface to any read-only consumer at all.

Item Retrieval

Section titled “Item Retrieval”

These tools fetch full product records once a specific item or item group has already been identified by ID, rather than by query.

  • 📦 Item Retrieval

    Retrieve the full record for a single item, for a batch of items by ID, or for every member of an item group.

Memory

Section titled “Memory”

These tools let an agent persist state across a conversation. All memory tools require Basic authentication.

  • 📝 Append Memory

    Append a batch of items to a conversation’s durable, append-only memory log.

  • ⏪ Recall Memory

    Retrieve the most recent items from a conversation’s memory log.

  • ⏯️ Replay Memory

    Retrieve items from a conversation’s memory log starting at a specific offset.

  • 📋 List Conversations

    List all conversations stored for a given site, with metadata.

  • 🧵 Summarize Conversation

    A condensed, up-to-date summary of a conversation so far. Documentation coming soon.

Shared Data Types

Section titled “Shared Data Types”

Response shapes common to several of these tools — McpItem, McpImage, McpSiblings, McpVaryingAttribute — are documented once in MCP Data Types rather than repeated on every tool page.