Learning DNA Profile

Problem

Personalized tutoring often resets every session: the model can answer questions but does not consistently remember the student's grade, weak spots, explanation preferences, or goals. Learning DNA Profile extracts student state from chat history into a readable and updatable internal record, serving as a foundation layer for other tutoring skills.

How it works

It follows a profile — retrieval — update loop:

• Initial profile: captures seed fields such as grade, most challenging subject, specific blocker, preferred explanation style, and one semester goal, then stores a marked first version.
• Daily retrieval: silently reads the profile at the start of a session, adjusts tone, difficulty, and entry angle, and links questions to known weak points when relevant.
• Continuous updates: responds to signals like “I got it,” “still confused,” repeated similar errors, test results, preference changes, and 30-day milestones to adjust mastery levels and review priority.
• Honesty checks: when a student appears to claim understanding, asks a short variant question to verify before changing the record.

Other skills should query the learning_dna context first before choosing a tutoring path.

Boundaries

This skill supports long-term learner profiles but does not replace question banks, assessment systems, or grade management. If student information is missing, the profile is stale, or the need is one-off Q&A, personalization is limited; it works best alongside subject-specific tutoring skills.