What it does
It works offline without a key for suggestions, state and question design, code generation and linting; only evaluation needs the key.
Agent tooling
An MCP server and agent skill for using System One at design time: decompose a judgment into questions, lint them, calibrate thresholds on labelled rows and rerank.
Only you can see your notes.
Screenshot unavailable. Open the experiment ↗
It works offline without a key for suggestions, state and question design, code generation and linting; only evaluation needs the key.
Your developer can add this tool to their AI coding assistant. The assistant uses it to design rules for your app. For example, it could write code to sort support tickets by topic and flag angry messages. The assistant drafts these rules offline.
Designing the rules and application code does not require an outside connection. Testing those rules on real examples requires an access key. This key connects your project to TypeSafe, an outside AI service, so it can evaluate your tests.