reskill

Extracts shared patterns from agent charters and histories into reusable skills.

1|Updated Jul 7, 2026
One-click install
npx skills add https://github.com/seiggy/maf-copilot-studio-demo --skill reskill-seiggy
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: reskill
Source: https://github.com/seiggy/maf-copilot-studio-demo/tree/main/.squad/templates/skills/reskill
Command: npx skills add https://github.com/seiggy/maf-copilot-studio-demo --skill reskill-seiggy

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Agent charters and history files accumulate duplicated boilerplate and repeated learnings over time, inflating per-agent context consumption. This Skill performs a team-wide optimization pass that audits, extracts, and trims that bloat into shared reusable skills. ## Core Features & Use Cases - Audit: Measures charter and history byte sizes and identifies boilerplate repeated across 3+ charters, shared domain knowledge, and mature learnings appearing 3+ times. - Extract: Creates or updates skills at .squad/skills/{skill-name}/SKILL.md with confidence levels (low, medium, high) based on observation frequency. - Trim: Reduces charters to a minimal template targeting ≤1.5KB and histories to ≤8KB, then reports before/after savings in a summary table. - Use Case: When a coordinator hears "team, reskill" or suspects charter bloat, run this pass to consolidate duplicated collaboration boilerplate into one shared skill and shrink every agent's context footprint. ## Quick Start Run a team-wide reskill pass to audit all agent charters and histories, extract shared patterns into skills, and report the context savings.

Frequently Asked Questions about reskill

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I reduce AI agent context window usage?

Audit agent charters and histories for repeated boilerplate, extract patterns shared across 3+ agents into reusable skills, and trim charters to a minimal template targeting 1.5KB or less. This removes duplicated content from every agent's loaded context.

When should repeated content be extracted into a shared skill?

Content appearing in only one charter stays in that charter. Content in two charters is a candidate if overlap exceeds 500 bytes. Content in three or more charters should always be extracted into a shared skill.

What should be kept in an agent charter after optimization?

Keep Identity, What I Own, unique How I Work patterns, domain-specific Boundaries, the tagline blockquote, and a single-line Model preference. Remove boilerplate collaboration sections, voice sections, and domain knowledge now covered by skills.

What are the limits of automated charter trimming?

Trimming must not delete unique per-agent identity, domain-specific knowledge, or the model preference line. It also must not touch decision logs or merge unrelated patterns into a single oversized skill.

How are skill confidence levels assigned during extraction?

Confidence is set to low for a first observation, medium when a pattern appears in two or more agents, and high when the pattern is team-wide. This signals how established each extracted skill is.