Publish your source
Use a dedicated repository with an identifiable maintainer, license, release history, and security contact. Do not bundle credentials or private data.
Community submissions
Build locally.
Share safely.
Develop your app in its own repository, publish an exact container image or npm package, and propose that version-pinned artifact for AI Lab review. Accepted apps ship in a new versioned release.
Four gates
A shareable app is a bounded app.
The review exists so another user can understand exactly what will run, where data lives, and what leaves their machine.
Name one artifact
Provide an official image digest. If upstream offers only version tags, include the exact tag and its resolved digest. Exact npm package versions are also accepted; latest and runtime installer scripts are rejected.
Declare the runtime
Use one loopback-only 3xxxx port, deterministic health checks, fixed named volumes, and no privileged mode, host networking, or container-runtime socket.
Submit reproducible evidence
Document install, browser result, restart, persistence, removal, architecture support, telemetry, network calls, and offline limits. Course contributors submit the repository URL as a proposal to 32dots/ai-lab.
No live marketplace feed. The public site cannot install an app, and the local launcher never executes user-supplied catalog data. Reviewers reproduce the evidence, pin the accepted artifact, and publish it only with a new AI Lab release.
Ready to propose
Bring the repository, not a mystery image.
Include the app identity, intended use, artifact reference, architectures, license, local port, health path, durable paths, required credentials, network behavior, and lifecycle checks.