Setup MCP to interact with AI Models

What is Model Context Protocol (MCP)?

Model Context Protocol (MCP) is an open protocol that defines how external tools, data sources, and applications can integrate with AI models in a structured and standardized way. It lets models discover, query, and interact with these external capabilities through a clear schema, instead of using brittle prompt hacks. This makes AI systems more extensible, secure, and easier to maintain across different environments.

How does it work?

The AI Model ask the MCP Server about the Services like Documents, Ressources and Tools. The MCP Server connects to Simplifier Instance and executes the AI Model Requests. MCP Server doesn’t run besides your Simplifier Instance; you have to setup it up locally or beside an AI Model.

MCP in Detail

Model Context Protocol offers three services categories to interact with an Large Language Model (LLM)

1. Tools

What they are: Declarative descriptions of actions or functions that a model (via the MCP client) can call.
Purpose: Expose capabilities of external systems in a structured way (e.g., “create a task,” “query a database,” “trigger a workflow”).
Behavior: Each tool includes a name, description, and a JSON schema describing its expected parameters and return structure.
Example: A tool called getCustomerDetails that accepts { “customerId”: “string” } and returns customer info from a CRM.

2. Documents

What they are: Read-only structured content that provides context to the model.
Purpose: Supply the model with relevant, possibly dynamic data (e.g., help articles, product descriptions, code snippets, or process definitions).
Behavior: The client can fetch documents from the MCP server, which indexes and exposes them in a standardized way, so the model doesn’t need to rely on scraping or brittle prompts.
Example: A “ProductCatalog” document that the model can reference when answering questions about available SKUs.

3. Resources

What they are: References to external systems, APIs, databases, or other structured data sources that tools and documents may depend on.
Purpose: Provide a uniform abstraction layer for where information or capabilities come from. Resources can be used by multiple tools/documents.
Behavior: Resources are described in a discoverable way (e.g., type, schema, location, credentials), but the model doesn’t access them directly — only through the MCP server.
Example: A REST API endpoint for a low-code platform, a PostgreSQL database, or a document store.

What do I need to get it working?

You have to install the Simplifier MCP Server on your Client and run it locally, besides an AI Chat or an IDE (Integrated Development Environment) like Visual Studio Code.

The actual Version of Simpifier MCP is released under

https://www.npmjs.com/package/@simplifierag/simplifier-mcp

You need also