1. The code checks the repository. Ajar reads the repository’s configuration on GitHub (permissions, workflows, approvals, protections) and works out what each role can reach. Fixed rules, no AI. That is the house scene and the access report.
2. AI reads only the access request. When someone asks for access, AI reads the text of the request and points out pressure patterns seen in past takeovers. It never analyzes the repository, and it talks about the request, never the person.
3. Three AIs vote without talking. Claude Opus 5.5, GPT-5.6 Sol and Gemini 4 Argon each read the request on their own, without seeing the others. The code counts the votes: a level or a signal is shown only when at least 2 agree; otherwise “no agreement”.
4. Here, or in your terminal. This site is the live demo for public repositories. In your terminal, ajar check OWNER/REPO runs on your own repository, private ones included, with your own GitHub login.
In the scene, the three characters together stand in for one collaborator with the role you chose. Doors are roles, the vault is credentials and releases, and “?” means the check couldn’t confirm something, so it is never treated as safe. The characters are illustrated guides, not AI agents and not a test run against the repository. The AI results on this site are pre-computed; no AI is called here. Ajar never changes repository settings.
