Devart Learn MCP Server

What is Devart Learn MCP Server?

Devart Learn MCP Server is Devart's official hosted MCP server that lets AI assistants search the documentation and knowledge sources for Devart products such as dbForge, dotConnect and dbExpress. It is open to anonymous clients and needs no sign-in or API key.

Devart makes database development tools, data connectivity components and integration products for SQL Server, MySQL, Oracle, PostgreSQL and many cloud data sources. Instead of switching to a browser and guessing search keywords, you ask a question in plain language and the assistant pulls the relevant passages from Devart's docs, each with its source URL, title and product name.

How to use Devart Learn MCP Server?

The server runs at a single public Streamable HTTP endpoint:

https://chatbot.devart.com/api/mcp

Devart notes that this URL is not a regular web page: opening it in a browser shows an error, which is expected. There is nothing to install and no credentials to configure.

Claude Code

claude mcp add --transport http devart-learn https://chatbot.devart.com/api/mcp

Claude Desktop and claude.ai

Devart Learn is listed in the Claude Connectors Directory. Open Settings > Connectors, search for "Devart Learn" and click Connect. Then enable it for a conversation from the "+" button next to the message box under Connectors. You can also add it as a custom connector with the URL above.

Cursor

Add an entry to ~/.cursor/mcp.json (or .cursor/mcp.json in a project):

{
  "mcpServers": {
    "devart-learn": {
      "url": "https://chatbot.devart.com/api/mcp"
    }
  }
}

VS Code

Add the server to .vscode/mcp.json:

{
  "servers": {
    "devart-learn": {
      "type": "http",
      "url": "https://chatbot.devart.com/api/mcp"
    }
  }
}

For other MCP clients, paste the same URL into the client's MCP or connector settings.

Key features of Devart Learn MCP Server

• Official server operated by Devart, pointing at its full product documentation.
• Semantic search: finds relevant content from the intent of a question, even when you don't know the exact term.
• Results are self-contained chunks that include the source URL, page title, product name, content type and Markdown content, so answers can cite the docs.
• Product discovery tool that returns the exact product names in the knowledge base, so searches can be scoped to one product.
• Covers a wide catalog: dbForge Studio and SQL Tools, dbForge Schema and Data Compare, dotConnect ADO.NET providers, dbExpress drivers, Excel Add-ins, Entity Developer, Code Compare and more.
• Anonymous, remote access with no API key, account or local install.

Devart Learn MCP Server tools

The server exposes two tools:

• get_devart_product_names returns the complete list of product names available in the knowledge base. The assistant calls it first when it needs a valid product name to narrow a search, for example "dbForge Studio for PostgreSQL" or "dotConnect for Oracle".
• search_devart_knowledge_sources runs a semantic search across the documentation and other knowledge sources and returns the most relevant chunks in descending order of relevance. If nothing relevant exists, the chunks may be weakly related, so the assistant should check the sources it gets back.

Use cases of Devart Learn MCP Server

• Answering how-to questions in the IDE: "How do I compare two SQL Server schemas and generate a sync script in dbForge Schema Compare?"
• Connection troubleshooting: "What connection string options does dotConnect for PostgreSQL support for SSL?"
• Picking the right product: "Which Devart products can work with Salesforce data?" The assistant can list product names and then search the relevant docs.
• Writing code against Devart components: ask for a sample that uses a dotConnect or dbExpress driver and have the assistant ground it in the official documentation.
• Support-style questions: licensing, activation or installation steps that are covered in Devart's knowledge sources.

Troubleshooting Devart Learn MCP Server

• Opening the URL in a browser shows an error. That is expected; the endpoint only answers MCP requests from a client.
• The assistant answers from memory instead of the docs. Make sure the connector is enabled for the current conversation (in Claude, toggle it on under Connectors), or ask explicitly to "search the Devart docs".
• Results look off-topic. Ask the assistant to call get_devart_product_names and restrict the search to the exact product you use; the search can return weakly related chunks when the docs don't cover a question.
• The client cannot connect. Check that your client supports remote Streamable HTTP servers and that the URL is entered exactly, including /api/mcp.