> ## Documentation Index
> Fetch the complete documentation index at: https://docs.tendy.by/llms.txt
> Use this file to discover all available pages before exploring further.

# Tendy MCP: Connect AI Assistants to Procurement Data

> Tendy's MCP server lets Claude, ChatGPT, and other AI assistants search live tenders, retrieve details, and run saved searches on your behalf.

The Tendy MCP server exposes your procurement data to AI assistants through the [Model Context Protocol](https://modelcontextprotocol.io), an open standard that lets large language models call external tools in a structured, secure way. Once connected, your AI assistant can search live tenders across all Belarusian procurement sources, pull full tender details, and work with your saved searches and bookmarks — all without you leaving your chat interface.

## What your AI assistant can do

* Search all Belarusian procurement sources by keyword, OKRB code, region, amount range, deadline urgency, and more
* Retrieve full tender details including lots, documents, requirements, and procuring-entity information
* List and execute your saved search templates to surface matching tenders on demand
* Access your bookmarked tenders to quickly review opportunities you've flagged

## Supported clients

You can connect any MCP-compatible AI assistant to Tendy. The following clients are verified or commonly used:

| Client                    | Auth method      |
| ------------------------- | ---------------- |
| Claude (Anthropic)        | OAuth 2.1 or PAT |
| ChatGPT (OpenAI)          | OAuth 2.1 or PAT |
| Mistral Le Chat           | PAT              |
| Grok (xAI)                | PAT              |
| Any MCP-compatible client | PAT              |

## Server details

| Property  | Value                      |
| --------- | -------------------------- |
| Endpoint  | `https://tendy.by/api/mcp` |
| Transport | Streamable HTTP            |
| SSE       | Disabled (stateless)       |

## Authentication

Tendy supports two authentication methods for MCP clients.

**Personal Access Token (PAT)** — Tokens prefixed `tndy_pat_…` are suitable for any MCP-compatible client. You generate a PAT in your account settings and pass it as a `Bearer` token in the `Authorization` header. Use this method for clients that require manual configuration.

**OAuth 2.1** — Claude and ChatGPT verified connectors support OAuth 2.1. These clients automatically discover the Tendy authorization server at `https://tendy.by/.well-known/oauth-protected-resource` and prompt you to authorize in-browser. Tendy issues an OAuth access token prefixed `tndy_at_…` — no manual token management is required.

<Note>
  MCP access requires the **Basic plan or above**. Upgrade your plan in Settings → Subscription if you are on a free tier.
</Note>

## Next steps

<CardGroup cols={2}>
  <Card title="Connect a Client" icon="plug" href="/integrations/mcp-connect">
    Generate a PAT or use OAuth to connect your AI assistant to Tendy in minutes.
  </Card>

  <Card title="MCP Tools Reference" icon="wrench" href="/integrations/mcp-tools">
    Explore all 5 MCP tools and their parameters, return values, and usage examples.
  </Card>
</CardGroup>