Skip to main content
SHARC publishes two packages to NPM for integration into your projects.

​
@sharc-code/mcp

npm version The MCP server that provides semantic code search to AI assistants.

​
Installation

npm install @sharc-code/mcp
# or
npx @sharc-code/mcp

​
Usage with Claude Code

claude mcp add sharc \
  -e SHARC_API_KEY=sk_mcp_xxx \
  -- npx @sharc-code/mcp

​
Features

  • 7 MCP tools for indexing and searching code
  • Real-time file watching for incremental updates
  • AST-based code chunking with context injection
  • Merkle-based sync for efficient re-indexing

​
@sharc-code/splitter

npm version Code chunking library with AST-based splitting and LangChain fallback.

​
Installation

npm install @sharc-code/splitter

​
Usage

import { AstCodeSplitter } from '@sharc-code/splitter';

const splitter = new AstCodeSplitter();

const code = `
export class UserService {
  async authenticate(token: string): Promise<User> {
    return jwt.verify(token);
  }
}
`;

const chunks = await splitter.split(code, 'typescript', 'services/user.ts');

console.log(chunks[0].content);
// Output:
// // Context: class UserService (services/user.ts)
// async authenticate(token: string): Promise<User> {
//   return jwt.verify(token);
// }

​
Supported Languages (AST)

LanguageExtensions
TypeScript.ts, .tsx
JavaScript.js, .jsx, .mjs, .cjs
Python.py
Go.go
Rust.rs
Java.java
C#.cs
C/C++.c, .cpp, .h
Scala.scala

​
Features

  • AST-based chunking: Extracts complete semantic units (functions, classes, methods)
  • Context injection: Adds parent class/module context to each chunk
  • Decorator support: Extracts @Decorator, #[attribute], and @Annotation context
  • LangChain fallback: Character-based chunking for unsupported languages
  • Syntax validation: Detects parse errors before chunking

​
API Reference

​
AstCodeSplitter

class AstCodeSplitter {
  constructor(chunkSize?: number, chunkOverlap?: number);
  
  async split(
    code: string,
    language: string,
    filePath?: string
  ): Promise<CodeChunk[]>;
  
  setChunkSize(size: number): void;
  setChunkOverlap(overlap: number): void;
  checkSyntaxErrors(code: string, language: string): SyntaxError[];
}

interface CodeChunk {
  content: string;
  startLine: number;
  endLine: number;
  metadata?: Record<string, any>;
}

​
LangChainCodeSplitter

class LangChainCodeSplitter {
  constructor(chunkSize?: number, chunkOverlap?: number);
  
  async split(
    code: string,
    language: string,
    filePath?: string
  ): Promise<CodeChunk[]>;
}

​
Package Versions

PackageVersionDescription
@sharc-code/mcp0.2.1MCP server for AI assistants
@sharc-code/splitter0.2.0AST and LangChain code splitters