A toolkit and visualizer for AI Bill of Materials (AI-BOM). Pipe any CycloneDX AI-BOM JSON into the aibom CLI and get an interactive constellation graph.
TL;DR how to use AI-BOM:
snyk aibom --experimental --json | npx aibom --viewWhat it does: Snyk generates a CycloneDX AI-BOM as JSON, which is piped into the aibom CLI. The --view flag opens an interactive HTML visualization of your AI bill of materials in the browser.
AI-powered systems are increasingly widespread, but understanding what's inside those models, including their components, data sources, dependencies, and risks—remains difficult. The AIBOM CLI helps developers by:
- Transparency: Instantly visualize the full "system composition" of your AI applications: AI models, datasets, libraries, and supply chain dependencies.
- Debugging & Operations: Find complex model dependencies to speed up troubleshooting and locate source-code usage of AI components in your AI/ML projects.
- Adoption with Existing Tools: Seamlessly integrate with tools like Snyk to generate and visualize AI-BOMs—no vendor lock-in, just pipe your JSON in.
The CLI turns complex JSON reports into an interactive constellation graph—making architectural risk, component drift, and dependency relationships easily explorable for all engineers involved in building, deploying, or reviewing AI-enabled software.
Watch a demo of the AIBOM CLI together with the Snyk CLI that generates the AIBOM payload:
aibom-demo-shorter.copy-2236x1238-20260308142954.mov
The AI-BOM web visualizer is deployed live here for public use: https://aibom.vercel.app
To visualize your AI-BOM in your own local environment, pipe a valid CycloneDX JSON data to the aibom npm CLI utility as follows:
cat data.json | npx aibom --viewYou can use the Snyk CLI (free) with the aibom command to create an AI-BOM and pipe it to the aibom npm package:
snyk aibom --experimental --json | npx aibom --viewInstead of --view, you can use the --serve flag to tell the aibom npm package to spin-up a local webserver and serve the HTML file:
npx aibom --serve --port 8081Please consult CONTRIBUTING for guidelines on contributing to this project.