reskill

Reduce team charter and history context overhead by extracting shared patterns and boilerplate into reusable skills, which trims unnecessary sections and lowers per-agent context usage. The process identifies recurring knowledge, promotes it to a skill, and generates a savings report.

Skill extraction from team histories is the process of identifying mature learnings and recurring patterns, then promoting them into standalone skills with assigned confidence levels based on occurrence frequency. This removes redundant text from the history while preserving the knowledge.

Yes, manual review is required before trimming team charters. The extraction identifies shared boilerplate and suggests trims, but you must validate and create the actual skills to ensure accuracy before context is permanently removed.

Yes, you can track context savings because the optimization process outputs a reporting table. This report details the exact reductions in charter and history sizes achieved by promoting recurring patterns to skills.

The best way to manage recurring boilerplate is to extract shared knowledge into reusable skills with confidence levels based on frequency. This actively trims the original histories to prevent context bloat while preserving team learnings.

You should avoid skill extraction for context reduction when team charters contain unique, non-recurring information. The process relies on identifying shared patterns and boilerplate to justify trimming histories and creating reusable skills.