/commit — Atomic commitsFour-pass methodology that catches the specific atomicity mistakes AI agents make — grouping by session, by shared prefix, or by proximity rather than by purpose.
/linear — Linear workflowFive commands covering the full development cycle, from planning through to merge — all without leaving Claude Code.
/craft — Prompt generationGenerate structured prompts using the C.R.A.F.T. framework — Context, Role, Action, Format, Target audience — that consistently produce exceptional LLM output.