airgap-check
Finds what an AI app still downloads or tries to reach when there is no network, before you deploy it offline.
- AI tool
- Free
Pricing
Free and open source (MIT).
Overview
airgap-check looks for the downloads and network calls an AI app still makes when it runs without internet access, such as tokenizer files, embedding weights, NLTK data or telemetry. A static scan reads your manifests and code for patterns like from_pretrained() or nltk.download(), and reads only the names of .env keys, never their values. With --run it starts your app in a Docker container with no network and logs every lookup and connection attempt, and --fix prints Dockerfile lines to pre-download files and turn off telemetry. You get a list of what would fail or leak offline and suggested fixes.
Key features
- Static scan of manifests and code patterns
- --run test in a Docker container with no network
- --fix prints offline Dockerfile settings
- Reads .env key names only, not values
- JSON output
- Go, no dependencies
Use cases
Show all 7Show fewer
Alternatives
Last updated 2026-11-10