Claude Code Indexer
🚀 Quick Install with VS Code
📖 Description
Multi-language code indexing with graph database, supports Python/JavaScript/TypeScript/Java/AutoIt, auto-ignores node_modules/.git, respects .gitignore, multi-keyword search, MCP for Claude Desktop, automated installation.
🎯 Features
- Multi-language Support: Python, JavaScript, TypeScript, Java, AutoIt
- Graph Database: Advanced code relationship mapping using Ensmallen
- Smart Ignoring: Automatically respects .gitignore and skips node_modules/.git
- Fast Search: Multi-keyword search with FTS5 full-text search
- MCP Integration: Seamless integration with Claude Desktop/VS Code
- Performance: ~50 files/sec with caching enabled
- AI Enhancement: Optional AI-powered metadata generation
📦 Installation
Via pip (Recommended)
pip install claude-code-indexer
Via VS Code MCP
Click one of the badges above to automatically configure Claude Code Indexer as an MCP server in VS Code.
Manual MCP Setup
# Install the package
pip install claude-code-indexer
# Setup MCP integration
cci mcp install
🚀 Quick Start
# Initialize project
cci init
# Index codebase
cci index .
# Search code
cci search "authentication"
# Find important components
cci query --important
# Get insights
cci insights
🤖 LLM Guidance (MCP-first)
If MCP is not set up, do this once:
cci mcp install
cci mcp-daemon start
When answering code questions, LLMs should always prefer MCP tools over manual file scanning:
index_codebase(if not indexed)search_codefor keywordsquery_important_codefor key componentsget_project_stats+get_codebase_insightsfor overviewenhance_metadataonly when needed
📊 Commands
Core Commands
cci init- Initialize project with CLAUDE.md and databasecci index [path]- Index source code into graph databasecci stats- Show project statistics
Query Commands
cci query [--important] [--type TYPE] [--limit N]- Query indexed entitiescci search <keywords>- Full-text search across codebasecci critical [--limit N]- Find most critical components
Analysis Commands
cci insights- Architectural analysis and health checkcci enhanced [filters]- Query AI-enhanced metadatacci enhance . [--limit N]- Add AI-powered metadata (requires API)
State Management
cci state capture- Save codebase snapshotcci state diff- Show changes since last capturecci state tasks- List development task history
MCP Integration
cci mcp install- Setup MCP server for Claudecci mcp-daemon [start|stop|status]- Background service management
🔧 MCP Server Configuration
When installed via VS Code badges or manual MCP setup, the server provides these tools:
Available MCP Tools
index_codebase- Index project codesearch_code- Search codebase with keywordsquery_important_code- Find critical componentsget_project_stats- Get project statisticsget_codebase_insights- Health and architecture analysisenhance_metadata- Add AI-powered metadatastore_llm_memory- Store analysis insightsstore_coding_pattern- Save reusable patternsstore_best_practice- Document best practices
📈 Performance
| Operation | First Run | With Cache | Notes |
|---|---|---|---|
| init | 1-2s | N/A | One-time |
| index | 5-10s | 0.5-1s | 95%+ cache hit |
| query | 0.1-0.5s | 0.05s | Database query |
| search | 0.2-1s | 0.1s | Full-text search |
🆚 Competitor Comparison (high level)
| Approach | Strengths | Tradeoffs |
|---|---|---|
| CCI (graph + MCP) | Local-first, structured graph, MCP tools for LLMs | Needs initial indexing and MCP setup |
| Sourcegraph/OpenGrok | Large-scale code search UI | Heavier setup, less LLM-native workflows |
| LSP/ctags | Fast editor navigation | Limited cross-file graph insights |
| Vector search only | Strong semantic matching | Weaker structural/relationship precision |
📊 Benchmarks (required for real comparisons)
We do not claim superiority without measurable data. Use the scripts below to capture real numbers on your machine and publish them.
# Cold run (no cache)
cci index . --force
# Warm run (cache)
cci index .
# Database benchmark
cci benchmark --records 1000
Document your results in docs/benchmarks.md.
⚠️ Security
- Never index sensitive data (API keys, tokens, credentials)
- Check for .env files:
find . -name "*.env" - Add sensitive files to .gitignore
🤝 Contributing
Issues and pull requests are welcome at GitHub.
- Contributing guide: CONTRIBUTING.md
- Code of Conduct: CODE_OF_CONDUCT.md
- Security policy: SECURITY.md
- Support: SUPPORT.md
- Index metrics: docs/index-metrics.md
📄 License
MIT License - see LICENSE file for details.
🔗 Links
Metadata
Release files for claude-code-indexer 1.29.1
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | |
|---|---|---|---|
| claude_code_indexer-1.29.1.tar.gz | 248.0 kB | Details |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Reset |
|---|---|---|---|---|
| claude_code_indexer-1.29.1-py3-none-any.whl | Python 3 | none | any | Details |
Total release size: 253.7 kB
Release files / claude_code_indexer-1.29.1.tar.gz
| Download URL | claude_code_indexer-1.29.1.tar.gz |
|---|---|
| Size | 248.0 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
286e89c7aaa7fe5e3b91a9d47021c5d75dfac80202cee52e4ef307112dae37e9
|
|
BLAKE2b-256 checksum How to use checksums |
bacf8fc479663bf7e7181a51b48647c03423656e109d388d0bb2f9a82c82c02e
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/6.1.0 CPython/3.13.2
|
Release files / claude_code_indexer-1.29.1-py3-none-any.whl
| Download URL | claude_code_indexer-1.29.1-py3-none-any.whl |
|---|---|
| Size | 5.7 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
603ef87f75481abcdab0055984c30063a369d0cc083d6d8a68c5468d69ee787a
|
|
BLAKE2b-256 checksum How to use checksums |
de34c8ee5bacfa2b0ec4153c8ae4c5dc7e6f6547722b20ae6caf537aff4f93c2
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/6.1.0 CPython/3.13.2
|