Connect the kinrg Model Context Protocol (MCP) server to your favorite AI Assistant to unlock semantic whole-codebase understanding.
Before configuring your IDE, you must install the kinrg CLI and build your project's Knowledge Graph.
Install globally via UV or Pip (supports specific versions):
# Install latest version from current directory
uv tool install .
# Or install specific version from pip (if published)
uv tool install kinrg@0.2.2
# Alternatively, using standard pip
pip install kinrg==0.2.2
Navigate to your project folder and run these essential commands:
cd /path/to/your/project
# 1. Build the Graph Database
kinrg build
# 2. Watch for file changes (Auto-updates graph when you save)
kinrg watch
# 3. View the graph in your browser
kinrg visualize
# 4. Integrate rapidly with known IDEs
kinrg install
.vscode folder if it doesn't exist.mcp.json.{
"mcpServers": {
"kinrg": {
"command": "/Users/hotienky/.local/bin/kinrg",
"args": [
"serve"
]
}
}
}
Cmd + ,).kinrgcommand/Users/hotienky/.local/bin/kinrg servecline_mcp_settings.json or ~/.continue/config.json:{
"mcpServers": {
"kinrg": {
"command": "/Users/hotienky/.local/bin/kinrg",
"args": ["serve"]
}
}
}
~/.codeium/windsurf/mcp_config.json.{
"mcpServers": {
"kinrg": {
"command": "/Users/hotienky/.local/bin/kinrg",
"args": ["serve"]
}
}
}
{
"mcpServers": {
"kinrg": {
"command": "/Users/hotienky/.local/bin/kinrg",
"args": ["serve"]
}
}
}
Cmd+Q) and restart. Click the 🔨 hammer icon to verify available tools!uvx: file not found in $PATH, it means your graphic application wasn't launched with your native bash/zsh shell PATH environment.
/Users/hotienky/.local/bin/kinrg to ensure bullet-proof execution across all IDEs. Refrain from just typing `uvx`.