Skip to content

Installation

One-liner (macOS / Linux)

curl -fsSL https://raw.githubusercontent.com/UmairBaig8/okf-generator/main/scripts/install.sh | bash

This installs the core package with all language parsers.

pip install okf-generator

Core extraction is fully offline and deterministic — no LLM, no API key, no internet required.

Optional extras

Extras What you get Command
[llm] LLM enrichment (OpenAI SDK) pip install "okf-generator[llm]"
[dashboard] Live bundle browser (FastAPI + uvicorn) pip install "okf-generator[dashboard]"
[dev] Development tools (pytest, ruff, mkdocs) pip install "okf-generator[dev]"

All at once: 

pip install "okf-generator[llm,dashboard]"

uv (fast alternative)

uv pip install okf-generator

Or with extras: 

uv pip install "okf-generator[llm,dashboard]"

Docker

docker pull ghcr.io/umairbaig8/okf-generator/okf-generator:latest

# Generate a bundle
docker run -v "$PWD:/repo" ghcr.io/umairbaig8/okf-generator/okf-generator \
  generate /repo/src /repo/okf_bundle

Verify installation

okf --version
# Should print: okf-generator v0.1.39

okf --help
# Shows all available commands

Quick smoke test

# Create a tiny test project
mkdir -p /tmp/test_project
cat > /tmp/test_project/greet.py << 'EOF'
"""A simple greeting module."""
def greet(name: str) -> str:
    """Return a friendly greeting."""
    return f"Hello, {name}!"
EOF

# Generate a knowledge bundle
okf generate /tmp/test_project /tmp/test_bundle

# Look up the concept
okf lookup --bundle /tmp/test_bundle greet

You should see a concept card for greet with its signature, docstring, parameters, and return type — all extracted without a single LLM call.

Platform-specific notes

Windows

Install via pip works on Windows, but some tree-sitter parsers require a C compiler. Use Windows Terminal + Python 3.11+:

pip install okf-generator

If you encounter build errors for language parsers, install Microsoft C++ Build Tools from visualstudio.microsoft.com.

Linux arm64 (Raspberry Pi, AWS Graviton)

All tree-sitter wheels are available for aarch64. Standard pip install works:

pip install okf-generator

Next steps