OpenGraph AI Docs

Getting started

Quickstart

Open repository

Build a knowledge graph from a folder of images, inspect the result, and ask your first question.

1. Get the repository

git clone https://github.com/OpenGraphAI/opengraph-ai.git
cd opengraph-ai/opengraph-image

2. Install the package

Create and activate your preferred Python environment, then install the package in editable mode:

pip install -e .

3. Configure a model provider

Copy the example environment file and add a supported model-provider key. For example:

cp ../.env.example .env
echo 'ANTHROPIC_API_KEY=your_anthropic_api_key_here' >> .env

Keep credentials local. Never commit .env or paste a real provider key into an issue, log, or prompt.

4. Build the graph

Point the CLI at a folder of images:

opengraph-image build ./tests/sample_images/test_photos

The command writes the graph beside the source folder:

tests/sample_images/test_photos/opengraph-out/graph.json

5. Inspect the result

opengraph-image summary ./tests/sample_images/test_photos

The summary gives you a quick view of what the extraction produced before you start asking questions.

6. Query the graph

opengraph-image query "What's happening in the park photos?" \
  --graph ./tests/sample_images/test_photos/opengraph-out/graph.json

The answer is generated from the graph built from your own source images.

What next?