Labs
Copy, paste, and run curl and JS. The labs read TYPESAFE_API_KEY from your environment only. Never paste a key into this site. The conceptual project labs do not touch product repos.
- PrimitiveChoice: pick one from a set
Think of a menu with fixed options. You give Jev the list, and it picks one. Your code then reads the choice, the probabilities, and the confidence.
- PrimitiveNoul: a yes/no answer with a probability
Ask a yes/no question. You get a number from 0 to 1. That number is the chance the answer is yes. It is like a gut check with odds attached.
- PrimitiveScore: rate on a rubric
Think of a 1-to-5 star rubric. You set ordered levels, and Jev rates against them. You get back a score, a legend, probabilities, and confidence.
- PatternConfidence gates: act, review, or escalate
The answer tells you what to do. Confidence tells you how sure you can be before you act. High confidence: act. Middle: have someone review. Low: escalate to a human.
- PatternFan-out: ask many questions at once
Like working down a checklist. You ask several small questions in one call. Then your code uses the answers it needs and ignores the rest.
- Project labIntent routing, AZMDR style (conceptual)
Sort each incoming request into a bucket. Then send it to plain code, a specialist step, or a human. This is a learning example only. Do not touch AZMDR repos.
- Project labNewsletter triage (conceptual)
Like sorting mail over a bin. One call asks three things: is it urgent (Noul), what is it about (Choice), and should you keep it (Noul). Example theme only.
- Project labApp intake moderation (conceptual)
Sort new submissions as they arrive. Only auto-accept when confidence is high enough.
- Project labFit Desk check-in judgment (conceptual)
Read a short check-in message. Judge whether the person is sticking with the plan. Example theme only. This maps to Fit Desk as an example. It is not a live integration.