MakotoCite is a research-writing skill for Claude. It reads your sources through six independent readers, verifies every quote, page number, and DOI against the actual document, and marks anything it can't source — instead of smoothing it over.
A model that reads a source once can't catch its own misreadings. MakotoCite runs several independent readings, surfaces where they disagree, and only lets verified claims into your draft.
An empirical extractor, a methods critic, a contrarian, a subject-matter expert, a PhD examiner, and a statistical reviewer each read your sources separately — then the pipeline debates and adjudicates where they disagree.
Exact quote, page, and DOI checked against the actual document, with a retraction check — compiled into a confidence-rated literature matrix. Low-confidence claims are flagged, never hidden.
A hostile reviewer pass critiques the finished draft for overclaims and citation stretch — so the toughest reading happens at your desk, not the editor's.
Sold on Gumroad with a 14-day money-back guarantee. Requires a Claude subscription — the skill runs inside Claude.
The complete skill for one researcher: all seven pipeline stages, every module, and a full worked example.
Everything in Personal, for your whole research group or lab.
Pick one, or let me take a project end to end — from raw export to a clean, self-updating report.
Messy exports in, analysis-ready data out. I fix formats, parse text into real numbers, standardize dates, consolidate stray categories, remove duplicates, and flag what was missing in the source — never guessed.
Interactive dashboards that turn a raw spreadsheet into answers at a glance — KPIs, trends, and filters your team can actually use. Built in Plotly, Shiny, or your existing BI tool.
Recurring reports that build themselves. Set it up once and get a clean, on-brand summary on the schedule you need — no more copy-pasting into slides every month.
Sample projects built to show the process, using a demonstration dataset.
A clean BI-style report — KPIs, trend, and breakdowns by category, segment, and region.
A live dashboard with sidebar filters that recompute every chart in real time.
From a messy export to an analysis-ready dataset, with every fix documented.
Wabi Studio takes its name from wabi — the Japanese aesthetic of quiet, understated simplicity; finding beauty in what's stripped back to only what matters. That's exactly the work: taking noisy, chaotic data and returning something calm, clear, and easy to read.
MakotoCite borrows a second idea — makoto (誠), sincerity: write only what the sources actually say, and say plainly when they don't.
Behind it is one person who codes and lives in data all day — fluent in Python and R, and happiest turning a spreadsheet nobody wants to open into something people actually use.
Questions about MakotoCite, a pilot for your lab, or a dataset that needs taming — tell me what you're working with and I'll take it from there.