Appearance
MCP Server
As part of the Relewise API, we offer Code Assistance via Model Context Protocol (MCP)
This is currently a Preview Feature!
MCP is a new and rapidly evolving protocol, so expect, as an early adopter, that AI might not get everything right and can make mistakes.
What is this?
Model Context Protocol (MCP) is a new protocol that enables Large Language Models (LLMs) to call remote tools and retrieve precise, domain-specific information they don't inherently possess. As an example, if you ask ChatGPT how to use the Relewise API, it does have some high-level knowledge but is not an expert, and this leads to hallucinations (i.e., it makes up code that does not exist).
The Relewise-MCP Server addresses this issue by providing tools that give intricate and up-to-date information about our API, enabling you to use AI to write Relewise integration code in C#, TypeScript, PHP, or Java.
Getting Started
Relewise ApS disclaims liability for any issues arising from MCP usage
It is not dangerous to call the MCP Server by itself, but how AI interprets the data is out of our control, so we strongly recommend that you only use it on code that is under source control
Step 1: Set up MCP in your IDE
Below details how to set up the MCP Server in your selected IDE (click to expand)
Visual Studio / Visual Studio Code
Add the following to your .vscode\mcp.json file
json
{
"servers": {
"Relewise-MCP": {
"url" : "https://mcp.relewise.com"
}
}
}(Both VS and VS Code recognize this file)
More details about MCP for Visual Studio and Visual Studio Code can be found here.
Cursor
Add the following to your .cursor\mcp.json file
json
{
"mcpServers": {
"Relewise-MCP": {
"url": "https://mcp.relewise.com"
}
}
}More details about MCP for Cursor
Other
Most other tools use the same format as Cursor, but please check the official documentation.
Step 2: Include AI Rules in your Repo (Optional but highly recommended)
We recommend that you include the following in your AI Rules of your repo (GitHub Instructions, Cursor Rules, Agents.md, etc.). See the documentation of your specific IDE on how and where you include such rules.
markdown
# Relewise API Rules (Model Context Protocol Tools)
- Never use your World Knowledge when working with the Relewise API (Always use the Relewise Tools instead)
- Always call the tool 'code_getting_started' before any other tool, as it gives you more context on how to use the tools properly.Core Concepts
To effectively use the Model Context Protocol (MCP) with your IDE and AI assistant, it's useful to understand the following core terms:
MCP Server: A remote server that exposes structured information to your IDE’s AI agent. Relewise provides such a server with detailed API knowledge.
MCP Tools: Specific API endpoints defined by the MCP standard. These tools allow the AI to retrieve accurate and real-time data about the Relewise API. For example, they can return valid request/response schemas or method usage examples.
Agent Mode: A mode in AI-assisted IDEs where the AI behaves more like an agent, capable of calling external MCP servers to fetch domain-specific knowledge. Tools are only used in Agent Mode, not in code-completion or inline editing modes.
Frequently Asked Questions
Q: What Programming Languages are supported?
A: C#, TypeScript/JavaScript, PHP, and Java
Q: Do I have to pay to use the Relewise MCP Server?
A: The MCP Server is free to use.
Q: I can't use MCP Tools in Visual Studio Code. What could be wrong?
A: MCP needs to be enabled in settings (if the setting is disabled, you might need to enable it on GitHub Organization level)
Q: I can't get the chat in my IDE to call the MCP. What should I do?
A: Please check that your IDE has been configured in your MCP server registrations and that your chat is in Agent Mode. (In general, MCP servers are only used by 'Agent Modes' and not in code-completion/Ask/Edit modes.) Also, remember to add the AI Rules in your AI instruction files (Getting Started: Step 2).