Finding Skill Opportunities

Overview

A skill opportunity is a recurring, multi-step procedure that involves judgment and keeps getting repeated by hand. Two independent bodies of evidence show where those live:

• The repo — git history and existing automation. Repeated commit types, files that always change together, release/migration keywords, half-written Makefile targets and runbook docs.
• The sessions — local Claude Code transcripts. What people actually asked for, the tool sequences they ran, and the files they touched.

They fail in opposite directions, which is why the default is both. Git sees only work that ended in a commit, so a ritual done by hand every week in a repo nobody commits to is invisible to it. Sessions see the ask but not the long arc, and they only cover the transcripts on this machine.

This skill scans both deterministically (via the bundled scripts) so you don't eyeball thousands of commits or sessions, then applies judgment to turn the raw signals into a ranked list of skill candidates with evidence.

Core principle: Let the scripts find the repetition; you decide which repetitions are worth a skill.

When to Use