Codebase Semantic Search

Semantic code search with CLI, Neovim plugin, and MCP server integration.

Features

• Semantic Search: Find code by meaning, not just text
• Multi-language: Python, C++ via Tree-sitter
• Flexible Backends: Local models or OpenAI-compatible APIs
• Neovim Integration: Built-in editor workflow
• MCP Server: AI assistant integration via Model Context Protocol
• Vector Database: PostgreSQL with pgvector
• Hierarchical Config: Global + project JSONC settings

Quick Start

# Install
python -m build && pip install .

# Setup PostgreSQL with pgvector
docker run --name codebase-indexing -e POSTGRES_HOST_AUTH_METHOD=trust -p 5439:5432 -d pgvector/pgvector:0.8.0-pg17
createdb -h 127.0.0.1 -p 5439 -U postgres codebase_indexing
psql -h 127.0.0.1 -p 5439 -U postgres -d codebase_indexing -f create_tables.sql -v dim=1024

# Index and search
codebase index -a "`ls`"
codebase search -q "your search query"

MCP Server

Integrates with AI assistants like Claude Code:

# Start MCP server
codebase-mcp

Configure in your AI assistant settings:

{
  "mcpServers": {
    "codebase": {
      "command": "codebase-mcp"
    }
  }
}

Use semantic search: Find authentication-related code in the codebase

Neovim Plugin

{
  "jiangyinzuo/codebase-semantic-search",
  dependencies = {"nvim-lua/plenary.nvim"},
}

• :Codebase - Open search panel
• <C-S> - Submit query

Configuration

Example ~/.config/codebase/config.jsonc:

{
  "pgvector": {
    "dbname": "codebase_indexing",
    "user": "postgres",
    "host": "127.0.0.1",
    "port": "5439"
  },
  "model_provider": "openai",
  "openai": {"url": "http://localhost:8000"},
  "model": "/path/to/local/model/"
}

Development

# Run tests
pytest

# Run with database
python run_integration_tests.py --with-db

# MCP server tests
pytest tests/test_mcp_server.py -v --asyncio-mode=auto

Project structure: src/codebase/ with CLI, indexing, search, MCP server, and model providers.

MIT License