symbols-mcp
Read, inspect and navigate codebase symbols by connecting to a Language Server
README
<div align="center">
Symbols MCP
Read, inspect and navigate through codebase symbols by connecting to a Language Server
</div>
Introduction
By connecting to a Language Server of choice this MCP server makes it easy and efficient for coding agents to explore and navigate the codebase and its dependencies. The server offers a minimal toolset intended to be simple to use and light on the model's context.
Available Tools
outline: returns an outline of code symbols in a file. Either compact or with a small code snippetinspect: returns docs, signature, declaration and implementation locations for a symbol. For local and third-party ones (e.g. npm, NuGet, ... )search: returns matching symbols across the codebasereferences: finds all references of a symbol across the codebaserename: renames all references of a symbol across the codebasediagnostics: returns active diagnostics in a given filecompletion: returns a list of contextual completions at a given locationlogs: returns Language Server own logs for troubleshooting
Installation
Quickstart (run command)
Use the run command to start the MCP server with the desired Language Server command defined inline.
npx -y "@p1va/symbols" run [run options] <lsp-cmd> [lsp args]
See below configurations for the Language Servers tested. Other stdio Language Servers should work too. For simplicity examples follow Claude Code schema.
<details> <summary> <picture> <img src="https://img.shields.io/badge/-3670A0?&logo=python&logoColor=ffdd54" valign="middle"> </picture> <b>Pyright</b> </summary>
Pyright
Installation
npm install -g pyright
Verify Installation
pyright-langserver should be available
Configuration
{
"mcpServers": {
"symbols": {
"command": "npx",
"args": [
"-y", "@p1va/symbols@latest", "run",
"-w", "optional/path/to/workspace",
"pyright-langserver", "--stdio"
]
}
}
}
ℹ️ If you'd rather avoid installing pyright globally and are fine with a slower start up, you can substitute
pyright-langserver --stdioin the JSON above withnpx -y -p pyright pyright-langserver --stdio
Troubleshooting
Virtual Env not found
If the logs tool output includes errors or the diagnostics tool only reports module import errors even when none appear in the IDE these might be signs of Pyright not detecting the virtual environment.
You can update your pyproject.toml to correctly point it to the virtual environment location.
[tool.pyright]
venvPath = "."
venv = ".venv"
</details>
<details>
<summary> <picture> <img src="https://img.shields.io/badge/-%23007ACC.svg?logo=typescript&logoColor=white" valign="middle"> </picture> <b>TypeScript</b> </summary>
TypeScript Language Server
Installation
npm install -g typescipt-language-server
Verify Installation
typescipt-language-server --version
Configuration
{
"mcpServers": {
"symbols": {
"command": "npx",
"args": [
"-y", "@p1va/symbols@latest", "run",
"-w", "optional/path/to/workspace",
"typescript-language-server", "--stdio"
],
"env": {
// Keep at least one code file open for search to work
"SYMBOLS_PRELOAD_FILES": "src/index.ts;",
"SYMBOLS_DIAGNOSTICS_STRATEGY": "push"
}
}
}
}
ℹ️ If you'd rather avoid installing typescript-language-server globally and are fine with a slower start up, you can substitute
typescript-language-server --stdioin the JSON above withnpx -y typescript-language-server --stdio
Troubleshooting
Search: No results
For the search functionality to work the TS Language Server needs to compute the codebase index and keep in memory.
This is done by keeping at least one code file open at any time. Use the SYMBOLS_PRELOAD_FILES="src/index.ts variable with paths to a few files.
</details>
<details>
<summary> <picture> <img src="https://img.shields.io/badge/-blueviolet?logo=dotnet" valign="middle"> </picture> <b>Roslyn</b> </summary>
Roslyn Language Server
Installation
The official C# Language Server is distributed over the VS IDE NuGet feed as a self-contained executable.
To facilitate download and extraction we use the dotnet CLI with a temporary project file named ServerDownload.csproj with the following content:
<Project Sdk="Microsoft.NET.Sdk">
<PropertyGroup>
<PackageNameBase>Microsoft.CodeAnalysis.LanguageServer</PackageNameBase>
<PackageVersion>5.0.0-1.25353.13</PackageVersion>
<RestorePackagesPath Condition=" '$(RestorePackagesPath)' == '' ">/tmp/lsp-download</RestorePackagesPath>
<ServerPath Condition=" '$(DownloadPath)' == '' ">./LspServer/</ServerPath>
<TargetFramework>net9.0</TargetFramework>
<DisableImplicitNuGetFallbackFolder>true</DisableImplicitNuGetFallbackFolder>
<AutomaticallyUseReferenceAssemblyPackages>false</AutomaticallyUseReferenceAssemblyPackages>
<RestoreSources>
https://pkgs.dev.azure.com/azure-public/vside/_packaging/vs-impl/nuget/v3/index.json
</RestoreSources>
</PropertyGroup>
<ItemGroup>
<PackageDownload Include="$(PackageNameBase).$(Platform)" version="[$(PackageVersion)]" />
</ItemGroup>
<Target Name="SimplifyPath" AfterTargets="Restore">
<PropertyGroup>
<PackageIdFolderName>$(PackageNameBase.ToLower()).$(Platform.ToLower())</PackageIdFolderName>
<PackageContentPath>$(RestorePackagesPath)/$(PackageIdFolderName)/$(PackageVersion)/content/LanguageServer/$(Platform)/</PackageContentPath>
</PropertyGroup>
<ItemGroup>
<ServerFiles Include="$(PackageContentPath)**/*" />
</ItemGroup>
<Copy SourceFiles="@(ServerFiles)" DestinationFolder="$(ServerPath)%(RecursiveDir)" />
<RemoveDir Directories="$(RestorePackagesPath)" />
</Target>
</Project>
We then pick the platform identifier matching the machine from this list:
win-x64, win-arm64, linux-x64, linux-arm64, linux-musl-x64, linux-musl-arm64, osx-x64, osx-arm64 or neutral
And finally restore the temporary project to trigger the download of the Language Server.
Adjust both RestorePackagesPath and ServerPath to work on your machine and keep track of the latter.
SYMBOLS_ROSLYN_PATH=$HOME/.csharp-lsp
dotnet restore ServerDownload.csproj \
/p:Platform=your-platform-id \
/p:RestorePackagesPath=/tmp/your/download/location \
/p:ServerPath=$SYMBOLS_ROSLYN_PATH
Verify Installation
To verify the outcome of the installation we run the command below
$SYMBOLS_ROSLYN_PATH/Microsoft.CodeAnalysis.LanguageServer --version
Configuration
{
"mcpServers": {
"symbols": {
"command": "npx",
"args": [
"-y", "@p1va/symbols@latest", "run",
"-w", "optional/path/to/workspace",
"dotnet", "$SYMBOLS_ROSLYN_PATH/Microsoft.CodeAnalysis.LanguageServer.dll",
"--logLevel=Information",
"--extensionLogDirectory=$SYMBOLS_ROSLYN_PATH/logs",
"--stdio"
],
"env": {
"SYMBOLS_WORKSPACE_LOADER": "roslyn",
// Adjust this to your installation path
"SYMBOLS_ROSLYN_PATH": "$HOME/.csharp-lsp",
}
}
}
}
ℹ️ Roslyn Language Server is also provided with the C# Dev Kit extension for VS Code however the launch command is a bit more complicated and changes each time the extension is updated. If wanting to try it i suggest trying the other modality (
config init*start) which brings a template for the launch command
Troubleshooting
Search: No Results Found
If search doesn't find results before a file was read for the first time it's possible to warm up by pre-loading a few files from different projects with
"SYMBOLS_PRELOAD_FILES": "src/Project/Program.cs"
Linux: Max Number of Inotify Instances Reached
If on Linux LSP logs suggest that the maximum number of inotify per user instances has been reached it's possible to increase it with a value greater than the actual
sudo sysctl fs.inotify.max_user_instances=512
This allows the Language Server to keep monitoring files in the Solution/Project
Additionally JetBrains has more details on this issue
</details>
<details>
<summary> <picture> <img src="https://img.shields.io/badge/-%2300599C.svg?logo=c%2B%2B&logoColor=white" valign="middle"> </picture> <b>Clang</b> </summary>
Clang for C/C++
Verify Installation
clangd --help is available
Configuration
{
"mcpServers": {
"symbols": {
"command": "npx",
"args": [
"-y", "@p1va/symbols@latest", "run",
"-w", "optional/path/to/workspace",
"clangd",
],
"env": {
"SYMBOLS_DIAGNOSTICS_STRATEGY": "push",
"SYMBOLS_PRELOAD_FILES": "path/to/file.cpp"
}
}
}
}
Troubleshooting
General Errors
Ensure either compile_commands.json is found in the working directory or provide its directory path with --compile-commands-dir=path/to/dir
Search: No Results Found
Index is generate when the first file is opened. To warm up is possible to pre load and keep a one or more files opened by providing a list in SYMBOLS_PRELOAD_FILES
</details>
<details>
<summary> <picture> <img src="https://img.shields.io/badge/-%2300ADD8.svg?logo=go&logoColor=white" valign="middle"> </picture> <b>Gopls</b> </summary>
Gopls
Installation
go install golang.org/x/tools/gopls@latest
Verify Installation
gopls version
Configuration
{
"mcpServers": {
"symbols": {
"command": "npx",
"args": [
"-y", "@p1va/symbols@latest", "run",
"-w", "optional/path/to/workspace",
"gopls"
],
"env": {
"SYMBOLS_DIAGNOSTICS_STRATEGY": "push",
// Adjust these to your machine
"GOPATH" : "$HOME/go",
"GOCACHE": "$HOME/.cache/go-build",
"GOMODCACHE": "$HOME/go/pkg/mod"
}
}
}
}
</details>
<details>
<summary> <picture> <img src="https://img.shields.io/badge/-%23000000.svg?logo=rust&logoColor=white" valign="middle"> </picture> <b>Rust-analyzer</b> </summary>
Rust-analyzer
Installation
rustup component add rust-analyzer
Verify Installation
rust-analyzer --version
Configuration
{
"mcpServers": {
"symbols": {
"command": "npx",
"args": [
"-y", "@p1va/symbols@latest", "run",
"-w", "optional/path/to/workspace",
"rust-analyzer"
]
}
}
}
</details>
<details>
<summary> <picture> <img src="https://img.shields.io/badge/-ED8B00?logo=openjdk&logoColor=white" valign="middle"> </picture> <b>Eclipse JDT</b> </summary>
Eclipse JDT Language Server
Installation
Follow installation instructions on the Project's GitHub README
Verify Installation
$SYMBOLS_JDTLS_PATH/bin/jdtls --help
Configuration
{
"mcpServers": {
"symbols": {
"command": "npx",
"args": [
"-y", "@p1va/symbols@latest", "run",
"-w", "optional/path/to/workspace",
"$SYMBOLS_JDTLS_PATH/bin/jdtls",
"-configuration", "$HOME/.cache/jdtls/config",
"-data", "$HOME/.cache/jdtls/workspace/$SYMBOLS_WORKSPACE_NAME"
],
"env": {
"SYMBOLS_DIAGNOSTICS_STRATEGY": "push",
"SYMBOLS_JDTLS_PATH": "$HOME/.java-lsp/jdtls",
}
}
}
}
</details>
Auto-detection (config & start commands)
If you prefer to keep your MCP config minimal and portable, it's possible to move Language Server definitions out to a dedicated file and have the MCP server read from there and auto-detect which Language Server to launch depending on the codebase.
<details>
<summary> 🆕 <b>1. <code>config init</code></b> </summary>
Initialize Config
User-wide
To create a user-wide config file run the following command
npx -y "@p1va/symbols@latest" config init --global
This will create a configuration file to be found at:
- On Linux:
~/.config/symbols-nodejs/language-servers.yaml - On MacOS:
~/Library/Preferences/symbols-nodejs/language-servers.yaml - On Windows:
%APPDATA%\symbols-nodejs\Config\language-servers.yaml
Workspace (Defaults to Current Directory)
To create a workspace config file run this instead
npx -y "@p1va/symbols@latest" config init -w path/to/workspace
- Will initialize
path/to/workspace/language-servers.yaml
npx -y "@p1va/symbols@latest" config init
- Will initialize
./language-servers.yaml
</details>
<details>
<summary> ⚙️ <b>2. <code>config show</code></b> </summary>
Tweak Config
Use your editor (here i use code) to tweak the generated config file and comment, uncomment or add new language servers.
code $(npx -y @p1va/symbols config path)
Show Active Config
Finally run the command in your workspace (e.g. where you launch Claude Code) to see that the changes are being applied
npx -y @p1va/symbols config show
npx -y @p1va/symbols config show -w path/to/workspace
npx -y @p1va/symbols config show -c path/to/config.yaml
</details>
<details>
<summary> ⚡️ <b>3. <code>start</code></b> </summary>
Update your MCP configuration with this MCP server. The first Language Server having workspace_files matching any of the files seen at the root of the workspace will be launched
{
"mcpServers": {
"symbols": {
"command": "npx",
"args": [
"-y", "@p1va/symbols@latest", "start",
// Defaults to current directory
"-w", "optional/path/to/workspace",
// Defaults to language-servers.yaml in current workspace
"-c", "optional/path/to/config.yaml",
]
}
}
}
</details>
Development
pnpm lintoutputs the lint violationspnpm lint:fixattempts to fix lint violationspnpm formatformats the codebasepnpm devstarts in development modepnpm buildruns the linter and buildpnpm startstarts the built artifactspnpm test:unitruns the unit testspnpm test:integration:{language id}runs the integration tests for a given language
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.