A Claude Skill that fixes the order.
Almost everything with more than three parts has an order, and almost nobody decides it. The order arrives by accident. It's the sequence the material was made in, or remembered in, or the one a template handed you.
Then the deck drags and nobody can say where. The meeting runs out of time before the decision. The page loses people at the same scroll depth every time. The talk was good and somehow didn't land.
Music and comedy solved this decades ago because the feedback is instant. A set that peaks too early dies in the room and everybody watching knows it. That produced a real craft for arranging things, and it works on anything a person moves through in time.
sequencing runs that craft on your deck, page, agenda, talk, curriculum, launch, or argument. Ask Claude to improve a deck and it'll polish all ten slides and never tell you that slide seven should be slide two and slide four shouldn't exist.
Two mechanisms, and everything else serves them.
Every item needs a job. In a set, no slot is neutral. Each one opens, builds, peaks, rests, plants, pays off, lands, or closes. Assign a job to every item you have. Whatever's left over is a cut candidate, and it's usually the reason the thing drags.
The swap test. For every adjacent pair, ask whether swapping them loses anything. If it doesn't, the order isn't doing work at that seam. A sequence where most pairs swap freely isn't a sequence, it's a list, and the audience feels that even when they can't name it.
Before either one it asks the question everything depends on: can they leave? A pitch deck presented in a room and the same file sitting in an inbox are not the same artifact, and confusing the two is why decks that kill live die on slide four in email.
It never adds material. If something structural is missing, it names the gap and stops.
git clone https://github.com/MichaelDoesMarketing/sequencing.git ~/.claude/skills/sequencingOr grab the packaged .skill file from releases.
"This deck drags and I can't figure out where."
"Here's my agenda for Thursday. We never get to the decision."
"I have 20 slides of material and 12 minutes."
Four modes. Reorder when the items exist and the order is wrong. Build when you have material and no order. The cut when you have too much. And the walkout, which finds the exact seam where you lose people and lists everything stranded behind it.
Not a made-up deck. The 2008 AirBed & Breakfast seed deck, the most celebrated pitch deck in startup history. Order verified across two independent sources (MyPitchDecks, Failory) because the whole finding depends on getting the order right.
THE READ
Non-captive. A seed deck lives in an inbox. The curve runs 3-2-3-4-3-4-2-3-2-2-3-4-4-3, which peaks at slide 6 and then spends four consecutive slides in the basement before climbing again.
THE WALKOUT
Slide 10. Competition (9) and Competitive Advantage (10) are two adjacent slides in the same mode at the same low energy, and slide 10 repeats slide 8's content almost exactly. Both say events, partners, PR.
The item responsible is slide 7, Business Model. One line, "we take a 10% commission," which resolves completely and promises nothing. Completion is an exit ramp.
What's stranded behind it: Team. Press. User Testimonials. And the ask.
That's the finding. The deck's entire proof stack, the three slides where somebody other than the founders says this is real, sits at positions 11, 12, and 13, behind the four dullest slides in the deck. In a non-captive artifact.
THE NEW ORDER
Product moves from 6 to 4. In an inbox, showing the thing working early buys everything after it. Press moves from 12 to 7. It's proof, and proof's job is creating readers, not rewarding them. Testimonials moves from 13 to 9, deliberately split from Press by Market Size so two social-proof slides aren't adjacent. Adoption Strategy and Competitive Advantage merge into one slide. Same three points, twice.
THE GAP
The deck ends on the ask. Land the ask, then close on something that belongs to them. Named, not written, because this skill doesn't add material.
And the honest part: it worked. They raised $600K. Which is the actual lesson. Sequencing is a separate axis from content quality, and this deck won on content while leaving its best proof buried. Most decks don't have content that good to fall back on.
SKILL.md the workflow, modes, the two mechanisms, rules
references/energy.md the curve, the five shapes that fail, contrast, rest
references/attention.md captive vs non-captive, finding the walkout, openers and endings
references/moves.md the eight jobs, callbacks, the swap test, cutting and merging
references/shapes.md standard running orders for decks, talks, agendas, curricula, launches
assets/running-order.md the worksheet
assets/walkout.md the diagnostic
Some sequences aren't about attention. A safety briefing, a legal disclosure, a set of instructions, a funeral service. Those have obligations that outrank engagement, and rearranging them for energy is malpractice. The skill asks what a sequence owes before deciding what it should do.
battle-tested writes the line. versus picks the fight. subtext reads what came back. sequencing puts it in order.
MIT. Use it, fork it, ship it.