Skip to main content
Glama

Tolstoy MCP

Connect any MCP client — Claude, ChatGPT, Cursor, Gemini, and more — to Tolstoy, the agentic platform for AI-native ecommerce brands. Run your shoppable video workspace right from chat: generate marketing content, manage your media library, build and publish shoppable widgets, browse your catalog, tag products, and track ad performance.

Tolstoy exposes its remote MCP servers over Streamable HTTP. The main Tolstoy server is secured with OAuth 2.1 + PKCE — no API keys to paste, your client runs the OAuth sign-in on first connect. Shopper is a separate public, no-auth marketplace server.

Server

Endpoint

What it does

Tolstoy

https://apilb.gotolstoy.com/mcp/v1/mcp

Your full shoppable-video workspace — generate and iterate on marketing videos and images, manage your media library, build and publish shoppable widgets, browse your product catalog, tag products, and review Meta ad performance.

Tolstoy Library

https://apilb.gotolstoy.com/mcp/v1/library/mcp

Existing media, product tags, catalog, and shoppable widgets. This scoped connection does not expose Studio generation or paid-ad tools.

Tolstoy Shopper

https://apilb.gotolstoy.com/mcp/v1/shopper/mcp

Shop across every brand store on Tolstoy — marketplace-wide product search, full product details, and virtual try-on. Public, no sign-in.

Start with one product

Follow the first-workflow guide to confirm your workspace, find existing videos for a product, and review its current shoppable placement. It includes separate Cursor and Claude Library setup paths, example prompts, and recovery steps. The Cursor plugin also includes the tolstoy-product-media skill for this workflow.

For reporting, missing integrations, or file-transfer problems, use the reporting and recovery guide. It explains which connection and widget IDs each reporting tool accepts, how to recover a missing Meta Ads connection, and how to upload through the browser.

Related MCP server: Nexbid

Tools

The main Tolstoy server exposes the following tools.

Studio — content generation

  • generate_studio_content — start a new Studio generation (image or video) from a prompt.

  • iterate_studio_content — continue and refine an existing Studio generation in the same session.

  • get_studio_session — check status and retrieve the finished media for a generation.

  • list_studio_folders — list your Studio projects and folders.

  • list_studio_chats — list the chats inside a Studio project or folder.

Media library

  • list_assets — list your most recent library assets (videos, images, AI drafts).

  • search_assets — search the library by name or creator handle.

  • get_asset — fetch full details for an asset, including tagged products and playlists.

  • update_asset — rename an asset or toggle its favorite status.

Shoppable widgets

  • list_widgets — list your onsite shoppable video widgets (Stories, Carousel, Spotlight, For You Feed, Tile, Collection Grid, Bubble Feed), live or draft.

  • get_widget — inspect a widget: type, live status, content selection, and the actual videos it shows (per-product for PDP-mode shoppable widgets).

  • create_widget — create a new shoppable widget from a template, with the platform's defaults.

  • update_widget — rename, publish/unpublish, set PDP mode, or change a widget's content selection.

Products & catalog

  • list_collections — list the catalog collections for a connected store.

  • search_products — find store products by title or query.

  • browse_products — browse catalog products and collections for a connected store.

  • tag_video_product — tag (or untag) products on a video so it becomes shoppable in product-tagged playlists and PDP widgets.

Stores

  • list_stores — list the connected stores on your account; widget, catalog, and product tools can target any of them.

Paid ads

  • list_ad_campaigns — list your Meta ad campaigns with real delivery status (effectiveStatus truth-telling).

  • get_ads_performance — raw Meta performance, including spend, CTR, actions, and action values, per campaign, ad set, or ad. Confirm the merchant's conversion event before calculating ROAS; do not add different action types together.

  • publish_to_meta_ads_library — push a library video or image into Meta Ads Manager, ready for ad creation (no campaign, no spend).

AI widget reporting

  • list_ai_widgets — find widgets built with the AI Widget Builder.

  • get_ai_widget_analytics — read one AI widget's saved metrics. It accepts a widgetId from list_ai_widgets; a storefront video widget's publishId from list_widgets does not work here.

Shopper (separate public marketplace server — no auth)

  • search_products — search products across every brand store on Tolstoy, one marketplace-wide catalog.

  • get_product — full product details: description, variants, price, media, the brand's videos, and buy/cart links.

  • try_on — open an interactive virtual try-on of a product (participating brands); the shopper uploads a photo and sees themselves wearing the actual item.

App-aware clients (ChatGPT, Claude) also get interactive views — a shoppable-widget card grid and asset previews — rendered inline in chat.

Connect

  1. In your MCP client, add a custom connector / remote MCP server.

  2. Paste the Tolstoy (or Shopper) endpoint above.

  3. Sign in with your Tolstoy account when prompted (OAuth). Shopper needs no sign-in.

Per-client setup (Claude, ChatGPT, Cursor, Gemini CLI, Codex, Perplexity, Goose, Cherry Studio, and more) is available in the Tolstoy platform under Settings → MCP.

Cursor Marketplace plugin

This repository includes a Cursor plugin manifest in .cursor-plugin/plugin.json. Install the repository as a local plugin while developing, or install Tolstoy from the Cursor Marketplace after the listing is approved. The plugin adds the OAuth-secured Tolstoy workspace server. The public Tolstoy Shopper server remains available through the endpoint and client configuration documented above.

Example client config

{
  "mcpServers": {
    "tolstoy": { "url": "https://apilb.gotolstoy.com/mcp/v1/mcp" },
    "tolstoy-shopper": { "url": "https://apilb.gotolstoy.com/mcp/v1/shopper/mcp" }
  }
}

Codex CLI

codex mcp add targets STDIO servers, so add the remote server to ~/.codex/config.toml and run the OAuth login:

[mcp_servers.tolstoy]
url = "https://apilb.gotolstoy.com/mcp/v1/mcp"

[mcp_servers.tolstoy-shopper]
url = "https://apilb.gotolstoy.com/mcp/v1/shopper/mcp"
codex mcp login tolstoy

Authentication

OAuth 2.1 with PKCE, backed by Amazon Cognito. Discovery via RFC 9728 protected-resource metadata at the server's /.well-known endpoints. Each connection is bound to the Tolstoy workspace you authorize with. Shopper is a public marketplace server and requires no authentication.

Privacy, terms, and support

Registry

Published in the official MCP Registry:

  • io.github.GoTolstoy/studio — the main Tolstoy server

  • io.github.GoTolstoy/shopper

The server.json for each is in registry/.

About

Tolstoy is the agentic platform for AI-native ecommerce brands, powering shoppable video for thousands of stores — https://www.gotolstoy.com

Related MCP Connectors

Related MCP Servers

  • A
    license
    D
    quality
    D
    maintenance
    Enables natural-language control of e-commerce operations including product management, order processing, inventory tracking, customer service, content generation, and advertising analytics through 15 integrated MCP tools. Provides a local-first commerce automation solution with SQLite storage and extensible channel adapters for end-to-end online store workflows.
    15
    3
    MIT
  • A
    license
    Not graded
    quality
    C
    maintenance
    Agentic commerce infrastructure for AI agents. MCP-native product discovery, contextual ad matching, and purchase facilitation with European privacy compliance (nDSG/GDPR).
    MIT
  • A
    license
    A
    quality
    C
    maintenance
    An all-in-one AI content creation platform with 30 tools for generating images, videos, music, and managing enterprise social media and CRM. Enables content generation, image editing, e-commerce features, and social media management directly from MCP-compatible AI assistants.
    43
    4
    Apache 2.0