Short, complete answers. We launched in 2026 after running the audit on our own product first; the questions below are the ones a paying skeptic should ask us. For the full mechanics, read the methodology.
Bullshit Meter is an evidence-backed audit for startup ideas and pitches. You give it a public GitHub repo or paste your pitch; it reads the actual artifact, checks every claim against your code and the live market, and returns a scored report with receipts for every finding.
The Sniff Test is free and needs no account: it extracts your claims and shows what a full audit would check. The Full Bullshit Report is $29 at launch and includes one Re-Meter credit. A Re-Meter (re-running the audit after you fix things) is $19 on its own.
The Bullshit Index (0 to 100) measures the gap between what you claim and what the evidence supports. The Build Signal (0 to 100) measures whether there is enough real signal to keep building. They are scored on separate rubrics and never averaged, because a strong product can have a bullshit story and a great story can have nothing under it.
No. Your report lives at a private unguessable link, is excluded from search engines, and nothing is published unless you share the link yourself. The sample reports on this site are deliberate showcase pages, not user reports.
The audit reads up to 25 files from the public repo you point it at, plus whatever text you paste. Secrets are redacted before anything is sent to a model, and your sources are used to produce your report, not to train models.
Not today. It reads public GitHub repos and anything you paste in as text. If the code is private, paste the parts you want challenged (a README, an architecture note, your pitch) and the audit will work from those plus the live market.
A few minutes, not seconds. The pipeline reads your files, runs live external research with counter-searches, scores two rubrics, and runs an adversarial critic pass. Anything promising an instant verdict is skipping the part that matters.
No. It is an eight-stage pipeline powered by Claude: claim extraction, evidence planning, artifact reading, external research with mandatory counter-searches, rubric scoring where the weighted math is computed in code, an adversarial critic that deletes unfair findings, and a verdict arbiter. Every finding ships with a source you can click.
That is the product working, not a defect, and a verdict is not refunded for hurting. If the audit itself failed (crashed, misread sources, audited the wrong thing), email meter@bullshitmeter.dev within 7 days for a re-run or refund, your pick. If you think the verdict is wrong, fix the thing, hit Re-Meter, and make us eat our words.
Yes, and it says so out loud. Every report carries a Confidence rating based on how much real evidence was available, lists what it could not verify instead of guessing, and includes one to three ways to know the verdict was wrong. It audits what you provide and what is public; it cannot interview your customers or see private usage data.