Sponsored by Deepsite.site

glyph

Created By
benmyles7 months ago
🔮 glyph – mcp server to map your codebase with symbol outlines
Content

glyph

A Model Context Protocol (MCP) server that extracts symbol outlines from your codebase using Tree-sitter's declarative query language. Gives LLM coding agents the context they need with clean, efficient symbol extraction. Can also be used as a standalone CLI tool.

Installation

macOS

Install Go:

$ brew install go

Install the latest version of glyph:

$ GOBIN=/usr/local/bin go install "github.com/benmyles/glyph@latest"

What it does

glyph takes a file path glob, recursively discovers matching code files, and parses them using Tree-sitter's powerful query language to generate accurate symbol outlines. You control the level of detail—from high-level structure to complete function signatures with visibility modifiers.

Why glyph?

LLM coding agents work best when they understand your code's structure. glyph bridges that gap by providing clean, multi-file symbol maps that serve as efficient context without overwhelming token limits.

Key Features

  • Declarative Query-Based Extraction - Uses Tree-sitter's query language for precise, maintainable symbol extraction
  • Glob-based file discovery - Point it at your code with familiar patterns
  • Language-agnostic architecture - Easy to add new languages with just query patterns
  • Configurable detail levels - Choose how much context you need
  • Multi-file support - Get a unified view across your entire project
  • MCP-native - Built for seamless integration with AI coding workflows
  • High performance - Optimized Tree-sitter queries for fast extraction

Supported Languages

  • Go - Functions, methods, types, structs, interfaces, constants, variables
  • Java - Classes, interfaces, methods, constructors, fields, enums, records, annotations
  • JavaScript/TypeScript - Functions, classes, methods, arrow functions, variables, interfaces, type aliases
  • Python - Functions, classes, decorated definitions, assignments
  • Easy to extend - Adding new languages requires only ~20 lines of query patterns

Architecture

glyph uses a modern, declarative approach:

  • Tree-sitter queries replace complex AST traversal logic
  • Language-specific query patterns define what symbols to extract
  • Unified extraction engine works across all languages
  • Modular design with separate concerns for queries, extraction, formatting, and file handling

Symbol Types

glyph uses concise symbol type names to minimize token usage:

  • func - Functions and function-like declarations
  • method - Class/struct methods
  • class - Classes
  • interface - Interfaces
  • struct - Structs (Go)
  • type - Type declarations (Go)
  • const - Constants
  • var - Variables
  • field - Class/struct fields
  • constructor - Constructors
  • enum - Enumerations
  • record - Records (Java)
  • annotation - Annotations (Java)
  • property - Properties (TypeScript)

Usage

Integration with AI Coding Assistants

Claude Code

Add glyph to Claude Code using the MCP command:

# Add glyph as a local MCP server (for your current project)
claude mcp add glyph /usr/local/bin/glyph mcp

To verify the installation:

# List all configured servers
claude mcp list

# Check glyph server details
claude mcp get glyph

You can also check the server status anytime within Claude Code using the /mcp command.

Cursor

Add glyph to your Cursor MCP configuration:

  1. Open Cursor settings (Cmd/Ctrl + ,)
  2. Search for "MCP" or navigate to Extensions → MCP
  3. Add glyph to your MCP servers:
{
  "mcpServers": {
    "glyph": {
      "command": "/usr/local/bin/glyph",
      "args": ["mcp"]
    }
  }
}

After configuration, you can use glyph through the MCP interface to extract symbol outlines from your codebase, helping the AI understand your code structure better.

MCP Server Mode (default)

Run glyph as an MCP server for integration with LLM coding agents:

$ glyph mcp

CLI Mode

Use glyph directly from the command line to extract symbols:

$ glyph cli '/path/to/project/*.go'
$ glyph cli -detail=minimal '/path/to/project/**/*.js'

Options:

  • -detail: Level of detail (minimal, standard, or full). Default is standard.

Note: All file patterns must be absolute paths.

Detail Levels

Minimal

Shows just symbol names and types with line numbers:

- func: main (line 15)
- struct: Server (line 5)
- method: Start (line 10)

Standard (default)

Shows signatures and declarations:

- func: func main()
- struct: type Server struct
- method: func (s *Server) Start() error

Full

Shows complete symbol definitions with code blocks:

- func (lines 15-20):

func main() { server := &Server{} server.Start() }

Development

$ git clone https://github.com/benmyles/glyph
$ cd glyph
$ make clean
$ make test
$ make build
$ make install DESTDIR=/usr/local/bin

Adding New Languages

Thanks to the query-based architecture, adding support for a new language is straightforward:

  1. Add language detection in file_utils.go (~2 lines)
  2. Add query patterns in queries.go (~10-20 lines)
  3. Add to query dispatcher in queries.go (~5 lines)

Example for adding Rust support:

// In file_utils.go
case ".rs":
    return rust.GetLanguage(), nil

// In queries.go
var rustQueries = map[string]string{
    "functions": `
        (function_item
            name: (identifier) @name
        ) @function
    `,
    "structs": `
        (struct_item
            name: (type_identifier) @name
        ) @struct
    `,
    // ... more patterns
}

Performance

  • Fast parsing with Tree-sitter's incremental parsing
  • Optimized queries for efficient symbol extraction
  • Minimal memory usage with streaming file processing
  • Parser reuse for better performance across multiple files
Recommend Servers
TraeBuild with Free GPT-4.1 & Claude 3.7. Fully MCP-Ready.
Jina AI MCP ToolsA Model Context Protocol (MCP) server that integrates with Jina AI Search Foundation APIs.
Serper MCP ServerA Serper MCP Server
Context7Context7 MCP Server -- Up-to-date code documentation for LLMs and AI code editors
DeepChatYour AI Partner on Desktop
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。
AiimagemultistyleA Model Context Protocol (MCP) server for image generation and manipulation using fal.ai's Stable Diffusion model.
BlenderBlenderMCP connects Blender to Claude AI through the Model Context Protocol (MCP), allowing Claude to directly interact with and control Blender. This integration enables prompt assisted 3D modeling, scene creation, and manipulation.
Zhipu Web SearchZhipu Web Search MCP Server is a search engine specifically designed for large models. It integrates four search engines, allowing users to flexibly compare and switch between them. Building upon the web crawling and ranking capabilities of traditional search engines, it enhances intent recognition capabilities, returning results more suitable for large model processing (such as webpage titles, URLs, summaries, site names, site icons, etc.). This helps AI applications achieve "dynamic knowledge acquisition" and "precise scenario adaptation" capabilities.
Tavily Mcp
Playwright McpPlaywright MCP server
Visual Studio Code - Open Source ("Code - OSS")Visual Studio Code
Amap Maps高德地图官方 MCP Server
Howtocook Mcp基于Anduin2017 / HowToCook (程序员在家做饭指南)的mcp server,帮你推荐菜谱、规划膳食,解决“今天吃什么“的世纪难题; Based on Anduin2017/HowToCook (Programmer's Guide to Cooking at Home), MCP Server helps you recommend recipes, plan meals, and solve the century old problem of "what to eat today"
MiniMax MCPOfficial MiniMax Model Context Protocol (MCP) server that enables interaction with powerful Text to Speech, image generation and video generation APIs.
WindsurfThe new purpose-built IDE to harness magic
ChatWiseThe second fastest AI chatbot™
MCP AdvisorMCP Advisor & Installation - Use the right MCP server for your needs
CursorThe AI Code Editor
TimeA Model Context Protocol server that provides time and timezone conversion capabilities. This server enables LLMs to get current time information and perform timezone conversions using IANA timezone names, with automatic system timezone detection.
EdgeOne Pages MCPAn MCP service designed for deploying HTML content to EdgeOne Pages and obtaining an accessible public URL.