I built a Claude Code skill that discovers which automations to build — by mining my own work

A developer built a self-improving skill for Claude Code that mines recent work history to discover which automations to build. The system scans git history, terminal sessions, and memory to identify repeated multi-step sequences, then packages high-confidence patterns into reusable tools. In testing, it surfaced a pattern repeated 50+ times that had not been abstracted, and correctly declined to package an already-existing workflow.

Most automation projects build an automation. This one builds a system that discovers which automations to build — by mining your own recent work — and then packages the high-confidence ones into reusable tools. It's a self-improving skill for Claude Code https://claude.com/claude-code . You feel it but rarely measure it: the same multi-step sequence done by hand for the third time. A focused bug-fix → branch → conventional commit → PR cycle. A QA-sweep dispatch. A release deploy. Each repetition is a signal that an abstraction is missing — but nobody systematically finds those signals. They're scattered across git history, terminal sessions, and memory. distill-workflows does distill-workflows is a Claude Code skill that turns that scattered signal into reusable tooling, in one pass: .jsonl — repeated bash command patterns The diagram above maps the flow left-to-right: Signal Sources → Scanner scan.sh → Distill Engine → Packaging → Outputs , with a A few design choices worth calling out: Run against its own repo, the scanner surfaced a pattern repeated 50+ times that nobody had abstracted: a single-concern Android/Kotlin fix shipped as a small PR with the same conventions every time branch naming, conventional commit scope, base branch, size limit, no AI attribution . The engine packaged it into an android-fix-pr skill — exactly the kind of convention layer the generic commit / create-pr tools don't encode. The higher-frequency QA-sweep pattern was also detected, but the engine correctly declined to package it: it already existed as a parameterized workflow. Knowing when not to build is half the value. SKILL.md procedure the agent follows. scan.sh signal miner — jq over transcripts, git log shape analysis, gps recall , artifact inventory — emitting one structured report. watchlist.md ledger for cross-session frequency. Built with Claude Code. The core idea generalizes well beyond one repo: point a discovery loop at your own work history, score what repeats, and let the high-confidence patterns become tools.