BumbleCrawl v1.0.3

• Topics in the link graph now come from the distinctive terms of each page (WDF*IDF) instead of the AI model: sharper topics such as product series or subjects, more pages in a topic, and no model download needed.
• Topics are named after the terms that set them apart. Folders show their own name, and hovering a folder shows its path and what it is about.
• The legend is a column on the right of the link graph that folds away. Folders open and close with their icon.
• The local AI model is now optional and only used for the check "Pages off the site's topic". Download or delete it under Settings › Local AI model.
• The Rendered page tab shows a page's screenshot again.
• API keys and logins that can't be read on this computer are no longer deleted when another one is saved.
• Scheduled crawls: the dialog now says that a schedule keeps the crawl configuration from when it was saved.

BumbleCrawl v1.0.2

• Improved performance and polish across the interface.

BumbleCrawl v1.0.1

• Pages about the same topic are now found by the distinctive terms of each page's whole main content (WDF*IDF), not just its beginning. It works without the AI model, also for command-line and scheduled crawls.
• New detail tab "Similar pages": the page's top terms and the pages most like it, with the terms they share.
• New columns in the Content tab: top terms, most similar page and topic similarity.
• Links to pages blocked by robots.txt no longer count as broken links. On sites that block their search page, this raises the health score.
• Comparing pages with the local AI model is faster: it now only checks how well each page fits the site's topic.

BumbleCrawl v1.0.0

• Command line: run a crawl without opening the window and save CSV or Excel exports, the PDF report and a summary into a folder, e.g. for scripts and build servers.
• Scheduled crawls can now save CSV or Excel exports (internal URLs, issues, links, broken links) into a folder after every run.
• New checks with the local AI model: pages about the same topic in different words, and pages far from the rest of the site's topic.
• JavaScript redirects are detected when rendering and count in redirect chains and loops.
• New checks for H1 headings that are too short and for images set as CSS backgrounds.
• Response codes: new filters for permanent, temporary, and meta refresh or JavaScript redirects.
• Crawl as GPTBot, OAI-SearchBot, ClaudeBot or PerplexityBot to see your site as AI crawlers do.
• XML sitemaps: optional change frequency and priority by click depth, and last-modified dates can be left out.
• Export › Broken links with their sources: the links to fix, with the page each one is on.
• View source shows the rendered HTML next to the original when JavaScript rendering is on.
• URL details only lists values that were measured for the URL.
• Help › User guide opens the new documentation at bumblecrawl.com/docs.

BumbleCrawl v0.1.2

• New look: the BumbleCrawl bee as app icon and logo, honey-gold accent colour.
• Settings now have sections on the left, with an Upgrade section that shows the Pro licences.
• The dialog when a Free crawl stops at 500 URLs is simpler: Continue or Upgrade.

BumbleCrawl v0.1.1

• When a Free crawl stops at 500 URLs, a dialog now says so, with how many URLs were left and how to get Pro.

BumbleCrawl v0.1.0

First release of BumbleCrawl