Nothing to hand? Load the — a subcommand-driven deploy CLI whose help text supports a full page — or the , a four-option image compressor where an honest page has four examples and the lint says so. Both replay a saved run for free. Or press to watch the lint fail a malformed page without any model call.
A tldr page is a format, and formats are checkable
The tldr-pages project holds thousands of pages to one shape: a lowercase # title, one or two > description lines, a required More information: <url>. line, then 5–8 examples — each a one-line description ending in a colon and a backticked command with {{placeholder}} values. Every one of those rules is measured here, in this tab, for free: paste any page and the lint tells you exactly which line breaks which rule, including the subtle placeholder conventions like {{[-m|--message]}} {{message_text}}.
The help text is the evidence, and the parser reads it first
Your pasted --help or man text is parsed locally into facts: every option with its short form, long form and argument; every subcommand; every URL. Those facts power a free skeleton page, and they power the grounding check: a flag that appears in an example but not in the help text is printed by name, whether you typed it or the model did. What you paste is what the page may claim — nothing else.
The metered pass is judgement, and it must show its work
The skeleton lists options in documented order; a good tldr page instead carries the 5–8 things people actually reach for, ordered by frequency, phrased as tasks. That ranking is what the AI pass adds. It returns strict JSON — the same engine renders it, the same lint re-checks it in public, and its coverage notes must say what it left out. Pricing is honest: a worst-case amount is reserved, only what the run uses is charged, and untouched examples replay a saved run for free. Pages are saved to your account and searchable by meaning.
- What does TLDR Studio cost?
- The whole deterministic half is free and runs in this tab with no account: the help-text parser, the skeleton generator, the format lint, the mechanical autofix, the styled preview, the before/after comparison and the .md and .json exports. Only the AI pass is metered — it shows the worst-case amount it will reserve before you press the button, charges only what the run actually uses, and the two bundled examples replay a saved run for nothing.
- What is a tldr page?
- The page format of the tldr-pages project: a lowercase # title, one or two > description lines, a required More information: <url>. line, then five to eight examples, each a one-line description ending in a colon followed by a backticked command with {{placeholder}} values. Pages live at pages/<platform>/<command>.md.
- Can the AI pass invent a flag that does not exist?
- It is instructed not to, and it is checked rather than trusted. Every option and subcommand an example uses is matched against the options, subcommands and URLs the parser extracted from your pasted help text; anything it cannot vouch for is printed by name next to the page as an unvouched flag or subcommand. The same check runs on a page you wrote yourself.
- Can I lint a page I already wrote, without generating anything?
- Yes — press Lint a page I already have in the workbench and paste it. Every format rule is measured with no model call and no sign-in. Paste the command's help text above as well and the grounding check runs too.
- Does it help me open the pull request?
- Yes. Once a page is in the editor the Ship it panel shows what stands between it and a merged contribution — the failing lint rules, the target path, whether the platform folder is right, and the reminder to check the page does not already exist — then opens GitHub's new-file editor on tldr-pages with the path and the page body already filled in, and copies the project's <command>: add page title convention for you.
- What happens when I run it a second time?
- The button changes to Improve this draft: the page in the editor travels with the request as current_page, along with everything the lint found wrong with it, so the pass keeps what is right and fixes the rest. The previous page is kept as a baseline, and Compare with the previous page reports exactly which examples were added, removed or changed.
A derived work of @github/create-tldr-page (MIT, github/awesome-copilot): its template, its placeholder conventions and its 5–8-examples-by-frequency rule are what this studio generates, measures and enforces. tldr-pages is a community project (tldr.sh); this app is not affiliated with it — check the project's current style guide before submitting a page upstream.