# Context+ — Semantic Code Intelligence MCP Server

> MCP server combining RAG, Tree-sitter AST parsing, and spectral clustering to turn large codebases into searchable feature graphs. 99% accuracy on monorepos. 1,700+ GitHub stars.

## Install

Merge the JSON below into your `.mcp.json`:

## Quick Use

Add to your `.mcp.json`:

```json
{
  "mcpServers": {
    "contextplus": {
      "command": "npx",
      "args": ["-y", "@contextplus/mcp-server"]
    }
  }
}
```

Restart Claude Code. Ask: "Explain the authentication architecture" or "Which modules depend on the database layer?"

---

## Intro

Context+ is an MCP server that combines RAG, Tree-sitter AST parsing, spectral clustering, and Obsidian-style linking to turn massive codebases into searchable, hierarchical feature graphs with 1,700+ GitHub stars. Designed for developers working on large monorepos where dumping files into context is impractical, Context+ gives your AI agent deep semantic understanding of code architecture — functions, classes, modules, and their relationships. Best for teams on 100K+ LOC codebases who need precise code intelligence. Works with: Claude Code, Cursor, any MCP client. Setup time: under 2 minutes.

---

## Core Capabilities

### Feature Graph
Context+ builds a hierarchical graph of your codebase:

```
Project
├── Auth Module
│   ├── login() → calls validateToken(), createSession()
│   ├── logout() → calls destroySession()
│   └── middleware → imports from Auth Module
├── API Layer
│   ├── /users → depends on Auth, Database
│   ├── /products → depends on Database, Cache
│   └── /orders → depends on Auth, Database, Payment
└── Database Layer
    ├── connection pool
    ├── migrations
    └── models
```

### Tree-sitter AST Parsing
Accurate code parsing across 50+ languages using Tree-sitter. Extracts function signatures, class hierarchies, import chains, and type definitions.

### Spectral Clustering
Groups related code into logical features automatically. Discovers modules and boundaries that may not match the file system structure.

### Obsidian-Style Linking
Navigate between code concepts using bidirectional links:
- "Show everything connected to the User model"
- "What calls the payment processing pipeline?"
- "Trace data flow from API request to database write"

### Semantic Search
Natural language queries against the code graph:

```
"Where is authentication handled?"
"Which functions modify the order state?"
"Find all database queries that join more than 3 tables"
```

### Key Stats
- 1,700+ GitHub stars
- 50+ languages via Tree-sitter
- Spectral clustering for feature discovery
- Sub-second query response
- Handles 500K+ LOC codebases

### FAQ

**Q: What is Context+?**
A: Context+ is an MCP server that builds a semantic knowledge graph of your codebase, giving AI agents deep understanding of code architecture, dependencies, and relationships.

**Q: Is Context+ free?**
A: Yes, fully open-source under MIT license.

**Q: How is Context+ different from Codebase Memory MCP?**
A: Codebase Memory focuses on token-efficient retrieval. Context+ focuses on semantic understanding — it clusters code into features and maps relationships using spectral analysis and AST parsing.

---

## Source & Thanks

> Created by [ForLoopCodes](https://github.com/ForLoopCodes). Licensed under MIT.
>
> [contextplus](https://github.com/ForLoopCodes/contextplus) — ⭐ 1,700+

Thanks for bringing deep semantic code intelligence to MCP.

---

<!-- ZH -->

## 快速使用

将以下配置添加到 `.mcp.json`:

```json
{
  "mcpServers": {
    "contextplus": {
      "command": "npx",
      "args": ["-y", "@contextplus/mcp-server"]
    }
  }
}
```

重启 Claude Code 即可使用。

---

## 简介

Context+ 是一个 MCP 服务器，结合 RAG、Tree-sitter AST 解析、谱聚类和双向链接，将大型代码库转化为可搜索的层次化特征图谱，GitHub 1,700+ stars。专为在大型 monorepo 上工作的开发者设计。适合 100K+ 行代码的项目。

---

## 来源与感谢

> Created by [ForLoopCodes](https://github.com/ForLoopCodes). Licensed under MIT.
>
> [contextplus](https://github.com/ForLoopCodes/contextplus) — ⭐ 1,700+

---
Source: https://tokrepo.com/en/workflows/bf876394-1dbf-4b0a-9d41-179ea6522616
Author: MCP Hub