shebe

Overview

Shebe is a high-performance local code search tool that uses BM25 algorithm instead of embeddings or GPU acceleration. It provides fast indexing (2k-12k files/sec) with 2ms query latency and full UTF-8 support. The project includes 14 MCP tools designed specifically for coding agents to perform tasks like finding references, searching code, and analyzing codebases. It's positioned as a complement to structural tools like Serena MCP, providing ranked results for large codebases with polyglot search capabilities.

Try asking AI

After installing, here are 5 things you can ask your AI assistant:

• find_references

  Find all references to a symbol in the codebase

  Try:Find all references to SubstitutionFormatterTry:Show me all usages of the login function
• search_code

  Search for code patterns across the repository

  Try:Search for access log formattingTry:Find all implementations of the authentication API
• index_repository

  Index a repository for code search

  Try:Index the ~/envoy repositoryTry:Prepare my project directory for searching
• find_file

  Find files by pattern in the repository

  Try:Find all test files related to authenticationTry:Show me all configuration files
• read_file

  Read the complete contents of a file

  Try:Show me the full content of utils.pyTry:Read the authentication module
• preview_chunk

  Preview a file with context around a specific match

  Try:Show me the context around the login functionTry:Preview the code around this error
• reindex_session

  Update an existing index with new files

  Try:Reindex my session to include new filesTry:Update the index with recent changes
• list_sessions

  List all available indexed repositories

  Try:Show me all my indexed repositoriesTry:List available sessions
• get_session_info

  Get information about a specific indexed repository

  Try:Show me details about the envoy indexTry:Tell me about my current session
• upgrade_session

  Upgrade an older session to the latest schema

  Try:Upgrade my old sessionTry:Fix schema version mismatch
• list_dir

  List files in a directory with matching pattern

  Try:Show me all test files in this directoryTry:List all Python files in the project
• delete_session

  Delete an indexed repository

  Try:Remove the envoy index from my sessionsTry:Delete my old session

Note: Tool names extracted from the table and text references, but detailed documentation is in a separate file not provided here.

Comparable tools

Installation

brew tap shebe-oss/tap
brew install shebe

export SHEBE_VERSION=v0.5.8
curl -LO "https://github.com/shebe-oss/shebe-releases/releases/download/${SHEBE_VERSION}/shebe-${SHEBE_VERSION}-linux-x86_64.tar.gz"
curl -LO "https://github.com/shebe-oss/shebe-releases/releases/download/${SHEBE_VERSION}/shebe-${SHEBE_VERSION}-linux-x86_64.tar.gz.sha256"
sha256sum -c shebe-${SHEBE_VERSION}-linux-x86_64.tar.gz.sha256
tar -xzf shebe-${SHEBE_VERSION}-linux-x86_64.tar.gz
sudo mv shebe shebe-mcp /usr/local/bin/

{
  "mcpServers": {
    "shebe": {
      "command": "shebe-mcp",
      "args": []
    }
  }
}

FAQ

What makes Shebe different from other code search tools?

Shebe uses BM25 algorithm instead of embeddings or GPU acceleration, making it extremely fast with low resource requirements. It provides ranked results and is optimized for large codebases with 2ms query latency.

Does Shebe support remote code repositories?

No, repositories must be cloned locally before indexing. This is a trade-off for speed and privacy, as it doesn't require network access during search operations.

Compare shebe with