autoskill

This skill analyzes coding sessions to extract durable preferences from corrections and approvals, then proposes targeted updates to Skills that were active during the session. It acts as a learning mechanism across sessions, ensuring Claude improves based on feedback.

The user triggers autoskill after a session where Skills were used. The skill detects signals, filters for quality, maps them to the relevant Skill files, and proposes minimal, reversible edits for review.

When to activate

Trigger on explicit requests:

• "autoskill", "learn from this session", "update skills from these corrections"
• "remember this pattern", "make sure you do X next time"

Do NOT activate for one-off corrections or when the user declines skill modifications.

Related: observe-work

autoskill is the explicit, on-demand pass: you ask it to learn from this session, and it proposes edits to the skills that were active. observe-work runs continuously and writes what it notices to a durable observation log for a later review. They are complements, not alternatives — observe-work is what catches the insight you would otherwise have lost by the time you thought to ask, and autoskill is what you reach for when you want the edits now.

If an observation log exists, check it for open observations naming the skills you are about to edit before proposing changes to them. An observation already written about a skill is evidence you would otherwise re-derive from a single session's signals, and re-deriving it usually produces a weaker version of the same rule.