Operator Skills

Skills built to run a real Claude Code setup. Full source, install steps, and detail: the repo.

• qq-audit: the quality check Claude can’t run on its own code
• qq-cap-resume: keep a live session visibly resuming itself through a cap
• qq-externalize: route work to free models, not your Claude subscription
• qq-go-afk-ham: max-effort autonomous campaigns for a big build
• qq-go-afk-lean: conserve your Claude usage as a repeatable discipline
• qq-go-afk-smart: a bounded punch list, delegated and verified overnight
• qq-infoviz: turn a dataset into a chart worth publishing
• qq-mailbox: two Claude Code sessions that just talk to each other
• qq-smart-next-move: the move that compounds what you just built
• qq-weekend-burn: put your unused Claude quota to work every week
• ww-agent-watchdog: verify another agent’s work without redoing it
• ww-dashboard-ux: a hard rulebook for dashboards that don’t lie
• ww-google-docs: make generated docs match your house style
• ww-mistake-ledger: close the loop on a mistake so it never repeats
• ww-notion-docs: stop guessing why your Notion content won’t render
• ww-notion-proposal-hardener: stop losing edits to Notion’s silent failures
• ww-overnight-runner: a real contract for working while you’re away
• ww-plan-audit: stress-test a plan before it becomes code
• ww-premortem: ask what breaks this before you build it
• ww-questioning: ask instead of guess, before it costs you
• ww-rule15: stop mistaking a rendered screen for a working feature
• ww-skill-auditor: know what a skill does before you install it
• ww-smart-quote-fixer: get quotes and apostrophes right
• ww-steelman: argue against your own plan before you build it
• ww-copy-sweep: catch AI-tell punctuation before it ever lands in a file