← Open sourceAgent verification

case-study-proof

Lints case studies for unsourced numbers, causal wording without evidence and missing dates, with a claims-to-evidence template.

Status
Live
Launched
Licence
MIT
Language
JavaScript
Run it

One command, copied from the repo.

Install case-study-proof: npx github:reidifydesign/case-study-proof check case-study.md

reidifydesign/case-study-proof

What it does

You write the case study in Markdown, keep the figures in an evidence file, and it tells you which sentences make a claim the evidence does not back. Every figure in the text ends with a reference such as [E1]. Each reference has an entry in evidence.json: the metric, the value or before and after, the source, the period it covers and the day you read it.

The check command reports unsourced numbers, numbers the cited evidence does not support (including wrong percentages worked out from before and after), causal wording such as "led to" or "drove" with nothing cited, figures with no date, and placeholders left in. The map command lists every figure and its evidence, and init writes a skeleton whose [[CONFIRM: ...]] slots fail the check until they are filled. It needs Node 18 or later, has no dependencies, and nothing leaves your machine. The examples use one fictional shop, and every figure in them is invented.

It does not check that a source is true: it checks that the text, the evidence entry and the dates agree, so a made-up evidence entry passes. It reads one sentence at a time, English and Markdown text, and will miss claims worded in ways it does not know. Figures written as words, such as nine out of ten, are not read.

Why it exists

Reidify writes case studies under one rule: every figure is measured and dated, and no claim says our work caused a result. On 2026-09-30 we looked for a tool that checks that rule on a case study and did not find one. The nearest were a tool that checks product marketing copy against a codebase, and general claim-checking agent skills.

This tool does the narrow thing. You write the case study in Markdown, keep the figures in an evidence file, and it tells you which sentences make a claim the evidence does not back.