infoharness

Your docs disagree with themselves.

Paste your help center address. In about two minutes you get a graded list of the pages that contradict, repeat or outdate each other, the ones that make your support bot answer wrong.

Free. No signup. Reads up to 300 pages.

No docs handy? Try
What a scan finds

/billing/refunds

You can request a full refund within 14 days of purchase.

These disagree

/faq/payments

Refunds are available within 30 days, no questions asked.

Same question, two answers. Your AI assistant quotes whichever page it finds first.

What every scan checks

Seven checks, each with its own list of exact pages and quotes, so you know what to open and what to change.

CheckWhat we flagLooks like
ContradictionsDifferent prices, limits, deadlines or steps for the same fact on different pages.14 days on one page, 30 days on another
Duplicate pagesTwo pages that say almost the same thing, so one will drift out of date.“Setup guide” and “Setup guide (copy)”
Repeated sectionsThe same paragraph pasted into several articles.A warning copied into 12 pages
Outdated contentPages that date themselves, or describe a version nobody else mentions.“Last updated: March 2019”
Broken linksLinks inside your docs that return an error./docs/pricing → 404
Thin pagesPlaceholders and near-empty pages that search still finds.“Coming soon.”
Orphan pagesPages listed in your sitemap that no other page links to./docs/data-retention

How it works

  1. Paste an address

    Your docs or help center URL. No account, no install, nothing to configure.

  2. We read the public pages

    The crawler follows your sitemap and links, obeys robots.txt and stays inside the docs folder you gave it.

  3. Fix what matters first

    You get a grade, every finding with quotes and page links, and a badge to show when it's clean.

Guides

Short reads on keeping documentation accurate.

All 8 guides

Questions

Does it read private or logged-in pages?

No. It only reads pages anyone can open without signing in. Documentation behind a login can't be scanned yet.

Will it work on my site?

It works on any documentation site or help center that serves its text as HTML, which covers most platforms. Sites that only draw their content with JavaScript show up as unreadable.

How are contradictions found?

By comparing the numbers, limits and dates that different pages state about the same thing. Each finding shows both quotes side by side so you can judge it in seconds. Read the guide.

How do I find broken links in my help center?

Paste the help center address into the scanner. It requests every link, skips sites that block automated checks, and lists each broken link with the pages and menu items that contain it. Broken link checker.

How do I find duplicate or outdated articles in a knowledge base?

Scan it. Pages are compared with each other to find near-duplicates, and checked for old dates, old version numbers and long-unmodified pages. Find duplicate and outdated pages.

Why does my AI chatbot give wrong answers?

Usually because the documentation it reads has conflicting or stale pages. The model repeats the page it retrieves, so fix the sources first. Read why and how to fix it.

Who can see my report?

Anyone who has the link. Reports of publicly scanned sites also appear in the public reports list and in search results, so teams can compare. Reports for sites monitored from an account stay private. To remove a report, email us from the site's domain.

What does it cost?

Nothing. Scanning, watching your sites on a schedule and alerts by email, Slack or webhook are all free. What is included.

Can I stop the crawler?

Yes. Add InfoharnessBot to your robots.txt and it will skip your whole site. How to block it.

Find out what your docs say to an AI assistant.

Free. No signup. Reads up to 300 pages.