Layer 3 · evidence
Behavioral measurement of AI capability growth. Not self-report.
Prompt decomposition
How well a practitioner breaks a problem down. Grounded in Bloom.
Failure response pattern
How they react to errors. Grounded in Bandura mastery experiences.
Metacognitive prompting
Do they seek understanding, not just output. Grounded in Marton & Säljö.
What you get
A behavioral report per cohort. Target: Cohen's κ ≥ 0.6 inter-rater reliability. Portable badges for practitioners who score well.
Open-source; validation study forthcoming with academic partners.