knowledge-graduation

Graduating recurring patterns means identifying stable, broadly applicable solutions in your context entries and promoting them into standalone local skill files. This prevents teams from re-deriving the same patterns across projects and accelerates onboarding.

Evaluate pattern graduation readiness by checking four criteria: stability, breadth, non-obviousness, and conciseness. Run this assessment during retrospectives or context reviews after reaching task-volume milestones like 10+ tasks to ensure broad applicability.

Run a graduation workflow during retrospectives, context reviews, and task-volume milestones, such as after completing 10 or more tasks. This timing ensures patterns have matured enough to demonstrate stability and broad applicability across different scenarios.

Create a canonical SKILL.md file under your skills directory and link it directly to the original context entry. This graduation process provides standardized documentation for evaluated patterns while maintaining traceability back to the source context.

The best way to manage context reviews is applying a standardized graduation workflow that assesses stability, breadth, non-obviousness, and conciseness. This systematically identifies mature patterns and promotes them into reusable local skills with traceability.

Graduating patterns too early risks promoting unstable or narrow solutions into reusable skills. Waiting until task-volume milestones ensures patterns satisfy the breadth and stability criteria, preventing the need to re-derive or deprecate immature skills.