reskill

To reduce charter bloat, run a maintenance pass that audits agent histories to identify boilerplate text and shared domain knowledge, then extracts recurring patterns into reusable skills. This structured workflow trims redundant context while preserving mature learnings.

The best way to extract recurring patterns is to audit charters for shared knowledge, create or update skill files under a dedicated skills directory using a standard template, and assign a confidence level to each extracted skill for scalable team reuse.

Pattern extraction improves context optimization by identifying mature learnings and boilerplate across agent histories, then promoting them into shared skills. This trims noisy per-agent context and replaces duplicated text with reusable, confidence-scored skills.

Yes, you can automate knowledge management by running a periodic reskill pass that audits all charters, identifies shared domain knowledge, and automatically creates or updates skill files to consolidate and promote recurring information for the entire team.

Yes, frontmatter with a name and description is required when creating or updating extracted skill files. This metadata ensures the newly promoted skills remain discoverable and properly documented within the team's shared knowledge structure.

You should avoid a skill extraction pass if your charters lack recurring patterns or shared domain knowledge, as the audit requires sufficient boilerplate or mature learnings to identify, extract, and promote into reusable skills effectively.