Self-hosting
Harness is built local-first: the screen models and your memory already run on your own device, inside your browser. The natural next step is letting you run the whole stack yourself, end to end, and that is where this is headed.
What is coming
Self-hosting will let you:
- Run the entire Harness app from source, on your own infrastructure.
- Bring your own inference (a local model or your own key) with nothing managed in between.
- Read the code that touches your screen and your memory, and verify it for yourself.
Because a self-host provides its own inference, the managed-inference credit system and the paid Resident plan will not apply. Where billing is not configured, the capabilities otherwise gated behind Resident stay open, so a self-host will run the full feature set on your own terms.
Not open yet
The source is not public today. Self-hosting is planned, not available. We would rather ship it properly than half-open it.
Follow along
Want to know the moment it lands? Follow the Harness organization on GitHub. That is where the repositories, the release notes, and the setup instructions will show up first.