2 releases (1 stable)
| new 1.0.1 | Feb 12, 2026 |
|---|---|
| 0.1.0 |
|
| 0.0.4 | Feb 11, 2026 |
#328 in Text processing
365KB
9K
SLoC
kdex
All your knowledge. One search. AI-ready.
Index your code, docs, notes, and wikis locally.
Let AI assistants search everything you know.
The Problem
You have knowledge everywhere:
- Code across dozens of repositories
- Notes in Obsidian, Logseq, or markdown files
- Docs scattered in wikis and READMEs
When you ask an AI assistant for help, it has no idea what's in YOUR files. You end up copy-pasting context manually.
The Solution
# Index everything
kdex index ~/code/my-project
kdex index ~/Documents/obsidian-vault
kdex index ~/wiki
# Search instantly
kdex search "how to deploy"
# AI assistants can now search your knowledge
kdex mcp # Start MCP server for Copilot, Claude, Ollama
kdex creates a local search index that AI tools can query. Your data never leaves your machine.
β¨ Features
| Feature | Description |
|---|---|
| π Instant Search | SQLite FTS5 gives sub-millisecond full-text search |
| π€ AI-Ready | MCP server for GitHub Copilot, Claude, Ollama |
| π Universal | Code repos, Obsidian vaults, wikis, any markdown |
| π Local-First | Your data stays on your machine. Always. |
| β‘ Fast | Index 100k files in seconds, search in milliseconds |
| π Remote Repos | Add GitHub repos by URLβauto-cloned, auto-synced |
| π¦ Portable | Export/import config for easy machine migration |
| π₯οΈ Interactive TUI | Full-screen interface with preview panel |
π Quickstart
# Install
cargo install kdex
# Index your project
kdex index .
# Add a GitHub repo
kdex add --remote owner/repo
# Search (just type your query!)
kdex "authentication"
# Launch interactive mode
kdex
That's it. Your knowledge, instantly searchable.
π€ AI Integration
kdex speaks MCP (Model Context Protocol), so AI assistants can search your indexed content directly.
GitHub Copilot CLI
Add to ~/.config/github-copilot/mcp.json:
{
"mcpServers": {
"kdex": { "command": "kdex", "args": ["mcp"] }
}
}
Claude Desktop
Add to ~/.config/claude/claude_desktop_config.json:
{
"mcpServers": {
"kdex": { "command": "kdex", "args": ["mcp"] }
}
}
Ollama (Local LLMs)
Pipe search results to your local model:
kdex search "error handling" --json | \
jq -r '.results[].snippet' | \
ollama run llama3 "Explain this code:"
See MCP Integration Guide for more details.
π What It Indexes
| Source | Examples |
|---|---|
| Code Repositories | Any git repo, monorepos, microservices |
| Knowledge Bases | Obsidian vaults, Logseq graphs, Dendron |
| Documentation | Markdown wikis, READMEs, technical docs |
| Config Files | YAML, TOML, JSON with searchable content |
kdex respects .gitignore and skips binary files automatically.
π» Usage
Interactive Mode (TUI)
kdex
Full-screen interface with:
- Real-time search as you type
- File preview panel (
Ctrl+P) - Repository management
- Keyboard-driven navigation
Command Line
kdex index <path> # Index a directory
kdex search <query> # Search indexed content
kdex search <query> --json # JSON output for scripting
kdex list # List indexed repositories
kdex remove <path> # Remove from index
kdex mcp # Start MCP server
Run kdex --help for all options.
π§ Installation
Quick Install (recommended)
curl -sSf https://urbanisierung.github.io/kdex/install.sh | sh
This downloads the latest binary and installs it to ~/.local/bin. No Rust required.
To update, re-run the same command.
From crates.io
cargo install kdex
From source
git clone https://github.com/urbanisierung/kdex.git
cd kdex
cargo install --path .
Pre-built binaries
Download from GitHub Releases.
Requirements
- Linux or macOS (Windows: use cargo install or pre-built binaries)
- For building from source: Rust 1.88+ (install via rustup)
π Documentation
- Features β Full feature list
- Documentation β Detailed usage guide
- MCP Integration β AI assistant setup
- Ollama Integration β Local LLM workflow
- Roadmap β What's coming next
π οΈ Development
make ci # Run full CI pipeline (Docker)
make ci-quick # Quick checks (format + clippy)
make test # Run tests
make build # Build debug binary
See CONTRIBUTING.md for development setup.
License
MIT Β© urbanisierung
Dependencies
~90MB
~1.5M SLoC