An MCP for interacting with the Onyx programming language's documentation.
Configuration
View docs{
"mcpServers": {
"elias-michaias-onyx_mcp": {
"command": "npx",
"args": [
"@onyxlang/mcp-server",
"bridge",
"--url",
"https://mcp.onyxlang.io"
],
"env": {
"DEBUG": "true",
"GITHUB_TOKEN": "ghp_XXXXXXXXXXXXXXXXXXXXXX",
"MAX_CRAWL_LIMIT": "50"
}
}
}
}You can access Onyx language documentation and code examples through a dedicated MCP server that offers fast, read-only search and query capabilities. It keeps data collection separate from querying, so you get up-to-date information without triggering crawling from your MCP client.
Connect your MCP client to an Onyx MCP Server to perform fast searches across Onyx documentation, GitHub examples, and Onyx code snippets. You can use a hosted, always-up-to-date server via NPX, or run a local server for development and testing. Either way you get a consistent query interface for finding docs, functions, and code examples.
Common usage patterns include:
- Connecting through NPX to a hosted bridge and querying the latest Onyx information.
- Running a local MCP server during development to test searches against your own crawled data.
- Bridging a local HTTP server to Claude Desktop or another MCP-compatible client for REST-style querying.
- Running the CLI crawlers yourself to populate local data without exposing crawling controls in the MCP interface.Prerequisites: Node.js and npm or npx installed on your system. Ensure you have network access to fetch packages when needed.
Option A: Quick access with NPX (no local install)
- This uses a hosted MCP server and requires no installation.Option B: Install locally (recommended for development) - Clone, install, and run the server locally to test crawling and querying against your own data.
The MCP interface is read-only for querying. Crawling and data population are performed via CLI commands and are not accessible through the MCP interface. This separation ensures safe, predictable query behavior while you control data collection.
Environment variables shown for configuration include a GitHub token for higher API rate limits and optional debug settings. Set these in your environment as needed.
You can connect to the MCP server in two primary ways: a hosted NPX bridge or a local/hosted bridge via a local HTTP server. Use the method that fits your development workflow and security requirements.
{
"mcpServers": {
"onyx": {
"command": "npx",
"args": ["@onyxlang/mcp-server", "bridge", "--url", "https://mcp.onyxlang.io"]
}
}
}If you encounter connectivity issues, verify that the bridge URL is reachable and that your environment permits outgoing connections to the MCP host. For development, use a local server and bridge to http://localhost:3002 to ensure the MCP client and bridge communicate correctly.
This MCP server provides read-only search and query capabilities across Onyx language knowledge. Crawling tools are available through the CLI but are not exposed via the MCP interface, preserving a clean separation between data collection and querying.
Search official Onyx documentation and related docs, guides, and references.
Search Onyx code examples across GitHub repositories by topic.
Extract Onyx function definitions from GitHub sources.
Extract Onyx struct definitions from GitHub sources.
List available Onyx-related GitHub repositories.
Unified search across all crawled data sources.
Execute Onyx code in a sandbox for testing and debugging.
Execute WebAssembly code and return outputs for testing.
Build Onyx code files using the Onyx build tool in a specified directory.
Build an Onyx package using the Onyx package build tool.