Workshop planning · browser-local cut tickets · 2026 · Solo — product framing, test-first packing logic, interaction design, accessibility, responsive verification, production deployment

Kerf Map

A proportional saw-table instrument that makes blade kerf occupy real space before the first cut.

The problem

Makers and small workshop operators often begin with a list of finished part lengths, estimate how many stock boards to buy, and then write a separate shop note. That translation can hide blade-width loss and leave cut order ambiguous. Kerf Map keeps the bounded planning loop causal: enter up to 48 repeated parts, choose one stock length and blade kerf, inspect a longest-first plan where each part, striped kerf, and offcut occupies proportional room, then copy the same board order as a plain-text ticket. It is deterministic planning guidance rather than a global cutting-stock optimizer or machine controller, and it explicitly requires physical verification at the saw. Inputs stay in the tab; there is no upload, account, persistence, paid API, tracking SDK, email, payment, advertising, or destructive operation. Production deployment dpl_Ch9Jq3VaGXktXnn4stfMCkrCpF6T reached READY before independent Foundry harness checks proved HTTPS markers and the seven-step configuration-to-copy interaction at the stable alias.

Architecture

Key decisions

01

Make kerf a visible object

Every part ends with a striped blade-width mark inside a proportional stock board. The visual and arithmetic share one state, so blade loss cannot disappear behind a summary number.

02

Prefer a legible heuristic over a fake optimum

Cuts are sorted longest-first and placed on the first board with room. The interface names that method and avoids claiming globally minimal waste; the result is deterministic, fast, and easy to inspect.

03

Keep the workshop claim narrow

Kerf Map assumes one stock length, one kerf, straight cuts, and one kerf charge per part. It does not model defects, end trim, grain, mixed stock, miters, clamping, calibration, or measurement error, and it tells operators to verify at the machine.

04

Use an asymmetric Configure bench

Warm drawing stock, a near-black saw table, chartreuse measurement marks, birch offcuts, rust errors, Impact and Arial Narrow signage, Georgia guidance, and Courier evidence labels create cut-list-to-stock-to-ticket causality without feature tiles, blur, a generic center stack, or dashboard data slop.

05

Constrain the release environment

The shared host exhausted threads during a 16-worker production build after compilation and typechecking. A checked-in two-worker Next.js cap produced the complete local build and the same static output on Vercel without weakening any test or release gate.

Metrics

6
pure kerf-planning tests passing
7
production golden-path steps passing
2
desktop and mobile viewports verified without overflow
0
production dependency vulnerabilities