Do not read a manual. Run one route: choose the job you need done, copy one command, and keep the check that tells you whether it worked.
Use one Skill first. You do not need to understand the other 55 before you get a useful result.
Get a problem hypothesis, authority line, evidence plan, smallest reversible bet, and a stop signal before a polished answer locks you into the wrong work.
Turn a topic or outline into an editable presentation with one narrative job per slide and a structural check you can rerun.
Review a change through failure-mode, maintainer, and trust-boundary lenses. Get a concrete BLOCK, CONCERNS, or CLEAN verdict.
Find the real entry points, ownership boundaries, and verification route before an agent starts changing code on guesses.
Define responsibility, inputs, tools, memory, handoffs, evaluation, and refusal boundaries before automation grows roots.
See a vague AI-research request become a bounded task with evidence and a stop signal.
OPEN THE DEMO →See the visual and structural difference between a slide that competes with itself and one that lands.
OPEN THE DEMO →See three review lenses surface a release risk that a smooth explanation does not fix.
OPEN THE DEMO →The same package also covers research, hooks, responsive UI, API review, content, diagrams, source reading, and local tools. Browse by job when your problem changes—not before.
DAYAN_TEST_HOME to a temporary directory first.