Nine task-agnostic meta-skills that operate one level up. Instead of doing the work, each emits the reusable structure the work follows — the prompt, the plan, the schema, the constitution, the map.
Every skill sits on one of two axes. The make skills take an idea to a shipped thing. The keep skills make sure it stays good — durable, honest, and improving on its own.
From a spark to a deliverable. These skills shape how a thing gets thought about, planned, designed, and assembled.
After it exists, keep it honest. These skills compound quality, set the bar, stress-test the work, and carry context forward.
Each is a skill you invoke by name. It returns the reusable scaffold a category of work follows — not a one-off answer. Open one to see the page and its explainer.
A reusable meta-prompt: how to think about a task category, not a worked example.
liveA meta-plan: the typed scaffold a fleet of agents executes, before the plan is written.
liveA medium-specific design schema, executed into an audited artifact plus the system that makes it.
liveA one-liner expanded into a full, research-grounded deliverable bundle.
liveThe compounding map: how each unit of work makes the next one easier, and where to invest first.
liveA tailored craft constitution plus an evidence-based Definition of Done.
liveAdversarial multi-lens review: prioritized findings, blind-spot probes, a reusable lesson.
liveA rigorous, cross-verified, source-cited brief with explicit confidence and reusable knowledge atoms.
liveA context strategy and droppable-in memory scaffold so work compounds across sessions.
The make skills get you to a shipped thing. The keep skills make that thing better and remember why. Review hands a lesson to the compounding map; the map feeds the context scaffold; so your next build starts from everything the last one learned instead of a blank page. See the full nine-step order on the operating loop →
/meta-suite:meta-review.$meta-review in Codex or /meta-review in Cursor and Grok, or just ask.