new·The score now tells you which way it movedA brain's exam only ever grows: its own material writes questions, and so does every question a real caller asked and did not get answered. The score is a percentage over that growing set, so a brain that learned more could post a smaller number — and this week three did. One of them answered two MORE questions than the week before and showed eighteen points less. Printed as a single percentage, that reads as decline to a reader and as punishment to anyone who contributes material.all news →
mozg.beta
Sign in

Building With AI Agents · all subjects

verification/exams

4 notes, read out of this brain and free to use. Each one was extracted from a source and is re-checked against its exam.

Exam regenerates from current goal

When a brain's goal changes, the exam that no longer matches the goal must be discarded and a new one generated. This prevents the brain from being graded on outdated knowledge targets. Implemented in lib/goal.ts.

Prevent infinite exam recursion: family search does not recurse beyond one level

A brain's exam can search its own family (parent + siblings). Search does not recurse into grandchildren. This prevents infinite loops and keeps exam scope bounded. Implemented in lib/families.ts.

Exam after every ingest shows % learned and missing materials

After each ingest, the exam (generated from the brain's goal) runs. It reports: % of goal covered (learned), and which materials are missing to reach 100%. This is how you know if the brain is actually learning.

Goal change invalidates exams that no longer match

When the goal text changes, any exam that was generated from the old goal must be discarded. The new goal generates a new exam. Otherwise the brain is graded on outdated knowledge. Implemented in lib/goal.ts.

Give your agent this brain