Chromia LSP MCP
Enables AI assistants to query and analyze Rell code via Language Server Protocol features like hover information, completions, diagnostics, and code actions. Automatically manages the Rell LSP server and provides resource-based access with real-time subscriptions.
README
LSP MCP Server for Rell
A Model Context Protocol (MCP) server that provides access to Rell Language Server Protocol (LSP) capabilities through AI assistants.
Documentation
Overview
The LSP MCP Server for Rell enables AI agents like Claude to query and analyze Rell code by providing programmatic access to LSP features. This is only used for AI agents that don't have LSP built-in like Cursor:
- Hover Information - Get type information, documentation, and contextual details about symbols
- Code Completions - Get completion suggestions based on current context
- Diagnostics - Get errors, warnings, and other diagnostic messages
- Code Actions - Get available refactorings and quick fixes
- Resource-Based Access - Access LSP features via URI-based resources with real-time subscriptions
The server automatically downloads and manages the Rell LSP server, eliminating manual setup steps.
Requirements
- Node.js 20 or later or Bun
Installation
The MCP server is published to npm as @chromia/chromia-lsp-mcp.
Claude Code
claude mcp add chromia-lsp -- npx -y @chromia/chromia-lsp-mcp
Copilot in VS Code
code --add-mcp "{\"name\":\"chromia-lsp\",\"command\":\"npx\",\"args\":[\"-y\",\"@chromia/chromia-lsp-mcp\"]}"
Cursor and other editors
Add the server to the editor's MCP config, which in Cursor's case is ~/.cursor/mcp.json:
{
"mcpServers": {
"chromia-lsp": {
"type": "stdio",
"command": "npx",
"args": ["-y", "@chromia/chromia-lsp-mcp"]
}
}
}
Some editors use servers instead of mcpServers; the entry itself is the same either way.
Bun and version pinning
To run on Bun, use "command": "bunx" and drop the -y argument.
Two versions can be pinned independently: the MCP server the usual npm way, and the Rell LSP it drives as a positional argument, as in "args": ["-y", "@chromia/chromia-lsp-mcp@0.0.4", "0.16.2"]. Without the latter, the newest published Rell LSP is downloaded and cached.
Building from source
Only needed to work on the server itself. See Setup & Development.
Features
MCP Tools
start_lsp- Start the LSP server with a specified root directory (required before using other tools)get_info_on_location- Get hover information at a specific location in a fileget_completions- Get completion suggestions at a specific location in a fileget_code_actions- Get code actions for a specific range in a fileopen_document- Open a file in the LSP server for analysissave_document- Save a file in the LSP server to refresh diagnosticsclose_document- Close a file in the LSP serverget_diagnostics- Get diagnostic messages (errors, warnings) for open filesrestart_lsp_server- Restart the LSP server without restarting the MCP serverset_log_level- Change the server's logging verbosity level at runtime
MCP Resources
lsp-diagnostics://resources for accessing diagnostic messages with real-time updates via subscriptionslsp-hover://resources for retrieving hover information at specific file locationslsp-completions://resources for getting code completion suggestions at specific positions
Usage
Your MCP client launches the server; you never run it by hand. Ask the assistant to work on Rell code and it drives the tools itself, starting with start_lsp, which needs a project root. If it picks the wrong directory, name the right one:
Start the Rell LSP server with root directory /path/to/my/dapp
The Rell language server is downloaded to ~/.chromia/lsp-mcp/ the first time it starts and reused afterwards.
Logging
The server reports what it is doing to the client as MCP log notifications. To make it more verbose, ask the assistant to set the log level to debug, or start it that way from the client config:
{
"mcpServers": {
"chromia-lsp": {
"command": "npx",
"args": ["-y", "@chromia/chromia-lsp-mcp"],
"env": { "LOG_LEVEL": "debug" }
}
}
}
In Claude Code, claude --mcp-debug additionally shows the raw traffic between client and server.
Troubleshooting
On supported platforms the server runs the Rell LSP with a self-contained runtime bundle downloaded to ~/.chromia/lsp-mcp/ and no local Java is involved. If no bundle exists for your platform or the chosen LSP version, it falls back to your own Java: JAVA_HOME first, then PATH (which java / where java on Windows). If startup fails asking for Java, check that java -version reports 21 or later. Editors launched from the desktop rather than a terminal do not inherit your shell's PATH, so a JDK installed through a version manager such as SDKMAN or jenv is often invisible to them; set JAVA_HOME, install a system-wide JDK, or start the client from a terminal.
Version 0.16.x not found in GitLab registry means the pinned Rell LSP version does not exist. Drop the version argument to take the latest release.
LSP server not started. Call start_lsp first with a root directory. is expected right after the client connects. Ask the assistant to start the LSP server on your project root.
If the language server itself misbehaves, delete ~/.chromia/lsp-mcp/; the runtime bundle or jar is downloaded again on the next start.
License
MIT License
Acknowledgments
Built on @Tritlo/lsp-mcp, the original LSP-over-MCP implementation.
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。