Small, honest tools for working with AI.

Check which model an API is really serving, take the machine tone out of a draft, and catch broken llms.txt and SKILL.md files before you ship. Everything is free, with no signup.

Verify a model

For when you pay for one model and suspect you get another.

Model Fingerprint Lab

Every model is bad at picking random numbers in its own way. Send it three short number prompts, paste the replies, and the lab ranks the closest matches in a reference bank. Run it against the official API and against a reseller or relay that claims the same model, then compare. Only the numbers leave your browser, and nothing is stored.

Fingerprint a model
reference bankmodel fingerprints
68
reference models to match against
prompts to send3
sent to the servernumbers only
storednothing

Write like yourself

For drafts that came out of a chatbot and sound like it.

AI Humanizer

Paste a draft. It marks the phrases that read as machine-made, rewrites them without changing what you said, and lists where a detail only you know would help. Pick work email, cover letter or LinkedIn post and the rewrite follows that format. Checks and rewrites use an AI model; 5 free rewrites a day.

Humanize text
our test set136 texts
AI texts marked75 / 86
human texts marked3 / 50
avg score after rewrite86 to 54
Measures writing patterns, not who wrote it.

Check before you ship

Find out if a file will work before anyone else does.

llms.txt Checker

Paste a URL or the file. You get 10 checks against the llmstxt.org spec, a score out of 10, and a Markdown report to paste into an issue.

Check a file
Tranco Top 5002026-10-07
18.3%
of real sites serve an llms.txt · 53 of 289
cybersecurity50%
enterprise SaaS47%
dev / cloud26%
consumer web3%

Claude Skill Validator

Paste a SKILL.md or upload the skill folder as a .zip. It checks the frontmatter, the name rules, and the description Claude reads to decide when your skill runs.

Validate a skill
spec limitsAgent Skills
allowed frontmatter keys6
name, max length64 chars
description, max length1,024 chars
compatibility, max length500 chars
An extra key like version: fails the upload. Put it under metadata.

Generate

Start from a working file instead of a blank page.

llms.txt Generator

Fill in your site name, a one-line summary and your key pages. It writes an llms.txt that follows the spec, ready to save at your site root.

Generate llms.txt

Decisions API Request Builder

Choose yes/no, pick-one or score, fill in the fields, and copy a request body for OpenAI's POST /v1/decisions. Then slide the estimator to see the monthly cost.

Build a request
price per 1M tokenssrc: OpenAI
input, either API$0.10
Decisions output$0
Responses output$0.50
gpt-6-luna. The Decisions API doesn't bill output tokens.

Guides & research

Model notes

Developer tools