Drill: Redesign the slop: take a generic SaaS landing page (gradient hero, 3 feature cards, emoji icons) and redesign it to avoid every anti-slop rule in claude-design. Target: same content, 10x less generic.
Interpretation: Seventh pass, first consumer domain — marathon training plans for recreational 3:30-5:00 marathoners, where gradient AI-coach heroes and streak badges are the category disease. Register: official race-manual documentation — paper white, Archivo + Newsreader + JetBrains Mono, hairline rules, one bib-red accent. Product shown working via checkable artifacts (week extract, pace bands, 16-week ledger, race splits) whose arithmetic reconciles across every section.
Taper builds your 16 weeks around one goal pace, then does the boring part: when work, flu, or a kid's birthday eats a weekend, the long runs move — your goal doesn't. PDFs you can print. Exports your watch already reads.
Every plan starts with a 5 K time trial, or two recent races if you have them. The trial sets your VDOT. VDOT sets five training paces. The paces set every workout on every page that follows.
If the trial says your marathon pace is 5:34/km, no amount of "how badly do you want it?" changes it. Wanting is not a training stimulus. The bands are the honest part — everything else in your plan is arithmetic on top of them.
| Band | Pace /km | Where it shows up |
|---|---|---|
| E · easy | 5:53 – 6:22 | Most of the week. Recovery days, warm-ups. |
| M · marathon | 5:34 | Long-run finishes. Race day, every km of it. |
| T · threshold | 5:04 | Thursday tempo — 5 km of it inside 9. |
| I · interval | 4:41800s → 3:45 / rep | Tuesday reps. Six of them, this week. |
| R · repetition | 4:20 | 200s, hills, strides. Sharpening only. |
From Mara's 21:40 trial — that's 4:20 /km. VDOT ≈ 42, bands rounded to :05. Check it: 5 × 4:20 = 21:40.
| Wk | Phase | Km | Long | Note |
|---|---|---|---|---|
| 1 | BASE | 40 | 14 | trial week — 21:40 5K on Sat |
| 2 | BASE | 45 | 16 | |
| 3 | BASE | 50 | 18 | |
| 4 | CUT | 36 | 12 | every 4th week easier |
| 5 | BUILD | 52 | 20 | |
| 6 | BUILD | 56 | 22 | |
| 7 | BUILD | 58 | 24 | |
| 8 | CUT | 40 | 14 | |
| 9 | BUILD | 60 | 26 | smoke week — intervals moved, km held |
| 10 | BUILD | 52 | 20 | |
| 11 | BUILD | 56 | 24 | the week shown above ↑ |
| 12 | PEAK | 62 | 28 | peak week |
| 13 | TAPER | 48 | 20 | volume down, intensity stays |
| 14 | TAPER | 40 | 16 | |
| 15 | TAPER | 30 | 12 | you will feel like running more. don't. |
| 16 | RACE | 46 | 42.2 | CIM, Dec 6 — 42.2 of it is the race |
| 16 WKS · 771.0 KM TOTAL · LONGEST 28.0 (WK 12) · RACE 42.2 | ||||
Cut weeks every 4th, like clockwork. Week 16 is 46 km — 42.2 of it is Sunday. Add the column: 771.0.
Not a feed that reveals tomorrow's workout at 6 a.m. The entire 16 weeks is a document you read once, print, and stick to the fridge. You should be able to point at week 9 and know what you'll be doing in November.
The shape is boring on purpose: three build cycles, a cut week after each, two-week-overlapping taper. Boring survives contact with October.
Band paces per 5 K with a cumulative clock, negative split by design: 1:58:30 at the half, 1:56:03 home. It's the same M-pace arithmetic from week 1 — nothing new is asked of your legs in mile 24.
Laminate it. Fold it behind your bib. The watch buzzes; the card is the law.
| Km | Pace /km | Band | Clock |
|---|---|---|---|
| 0 – 5 | 5:40 | 28:20 | 28:20 |
| 5 – 10 | 5:38 | 28:10 | 56:30 |
| 10 – 15 | 5:36 | 28:00 | 1:24:30 |
| 15 – 20 | 5:35 | 27:55 | 1:52:25 |
| HALF · 21.1 KM — 1:58:30 (5:36.6 /KM) | |||
| 20 – 25 | 5:33 | 27:45 | 2:20:10 |
| 25 – 30 | 5:32 | 27:40 | 2:47:50 |
| 30 – 35 | 5:30 | 27:30 | 3:15:20 |
| 35 – 40 | 5:28 | 27:20 | 3:42:40 |
| 40 – 42.195 | 5:25 | 11:53 | 3:54:33 |
| FINISH | avg 5:33.6 | — | 3:54:33 |
Bands sum to 3:54:33 — 27 s under the 3:55:00 line. Check it.
Resume at the last completed week's volume. Intervals drop out first; the long run comes back before the tempo does. Your goal holds until you've missed two long runs inside the peak block — then we talk.
The long run moves to Monday before it shrinks. If Monday's gone too, cut it 30% and keep the goal-pace finish. Intervals are the cheapest thing to skip — 40 minutes of threshold keeps most of the fitness.
Cross-train the same minutes at the equivalent effort. If day 7 still hurts, the plan sends you the physio referral list for your city — not a pep talk.
Every training block in history has been interrupted. Most apps handle it by guilting you back into a streak. Taper handles it by rewriting the remaining weeks around what you actually ran — the same five paces, re-laid across the days you have.
Three rules cover 90% of bad weeks. They're in the plan, on paper, before anything goes wrong.
“I missed all of week nine — flu, then a work trip. The plan moved my long runs instead of guilting me about the intervals I skipped. Ran 3:58, still not in a streak”
A year is 9 × 12 = $108. We charge $79. Save $29 — check it.
Build my planCancel any time. Your PDFs and exports keep working — they're files, not features.