Grading runs the suite and checks each acceptance criterion, names what is missing without fixing it, and offers the next rung of a hint ladder whose every escalation is logged. The learning record and the concept ledger update from what actually happened, and the agent's posture holds throughout: while a workbook is open it explains, points, reviews and grades, but does not write the implementation or the test.
Meta
- feature: docs/features/roadmap-driven-learning
- estimate: M
- candidate stories: A slice is graded against its acceptance criteria without the diff being edited; The four-rung hint ladder escalates only on request and logs each escalation; A hinted concept returns to the drill queue early; The posture rule refuses to write the implementation or the test while a workbook is open; Learning records and the concept ledger update from the graded outcome; The completion test requires cold retrieval at a gap, not introduction alone
- source: decomposition of "A teaching stage that turns a planned roadmap into hands-on exercises, placing the required theory immediately before each one — /nxs.teach, producing a workbook" (2026-08-31)
Grading runs the suite and checks each acceptance criterion, names what is missing without fixing it, and offers the next rung of a hint ladder whose every escalation is logged. The learning record and the concept ledger update from what actually happened, and the agent's posture holds throughout: while a workbook is open it explains, points, reviews and grades, but does not write the implementation or the test.
Meta