Lsp MCP

🚀 LSP-MCP Server

LSP-MCP is a universal server that combines the Language Server Protocol (LSP) and the Model Context Protocol (MCP). It enables querying language models through a consistent interface and supports various programming languages and code-related tasks such as error detection, intelligent completion, and documentation generation.

🚀 Quick Start

Supported Languages

• Haskell
• TypeScript/JavaScript
• Python
• Java
• C/C++

📦 Installation

The LSP-MCP server can be installed and run using the following command:

npx tritlo/lsp-mcp [LANGUAGE] [LSP EXECUTABLE] [MCP ENDPOINT]

Parameter Explanation

• [LANGUAGE]: The language to run (e.g., haskell, typescript, etc.)
• [LSP EXECUTABLE]: The path to the LSP executable for the corresponding language
• [MCP ENDPOINT]: The URL of the MCP model endpoint

Example

To run the Haskell language server:

npx tritlo/lsp-mcp haskell /path/to/haskell-language-server-wrapper lsp

✨ Features

The LSP-MCP server provides the following core features:

1. Language Model Integration: Supports invoking multiple language models (e.g., GPT - 4) through the MCP protocol.
2. Code Analysis: Provides LSP functions such as error detection, intelligent completion, and documentation generation.
3. Extension System: Supports language - specific extensions to enhance functionality.

📚 Documentation

Tool Definitions

LSP-MCP defines the following tools:

1. Get Diagnostics (get_diagnostics)

Retrieve diagnostic information (errors, warnings, and hints) for a specified file.

Request Format

{
  "type": "request",
  "method": "get_diagnostics",
  "params": {
    "file_path": "/path/to/file",
    "language_id": "haskell"
  }
}

Response Format

[
  {
    "range": {
      "start": { "line": 1, "character": 0 },
      "end": { "line": 1, "character": 5 }
    },
    "severity": 1,
    "message": "Error message",
    "code": "ErrorCode"
  },
  ...
]

2. Get Info on Location (get_info_on_location)

Retrieve detailed information (such as type, symbol resolution, etc.) at a specified location in the code.

Request Format

{
  "type": "request",
  "method": "get_info_on_location",
  "params": {
    "file_path": "/path/to/file",
    "language_id": "haskell",
    "line": 42,
    "column": 10
  }
}

3. Get Completions (get_completions)

Retrieve code completion suggestions at a specified location.

Request Format

{
  "type": "request",
  "method": "get_completions",
  "params": {
    "file_path": "/path/to/file",

(Note: The original README seems to be incomplete for the get_completions request format. The above output keeps it as it is.)