AI Tutor plan sourcing, fallback, and program-switch policy

DEC-0025 - AI Tutor plan sourcing, fallback, and program-switch policy

Context

After confirming 7-day plan structure and cadence, plan-content selection and edge handling must be fixed to avoid ambiguity and complexity.

Decision

Required exercise sourcing:

• Daily required slot prioritizes weakest skill in learner’s current program.

Low-inventory fallback:

• If required candidate is insufficient, use existing similarity ladder A -> B -> C.
• Keep fallback inside current program boundary.

Program switch during active cycle:

• If learner changes program during active 7-day cycle, keep existing plan until cycle end.
• Regenerate next cycle using newly selected program.

Simplicity posture:

• Do not add extra branching or user-choice forks for these cases in V2 baseline.

Rationale

Weak-skill-first keeps plan focused on recovery value. Reusing the established ladder avoids duplicate logic. End-of-cycle switch handling reduces mid-cycle churn and keeps behavior predictable for learners.