Anyone up for testing my new AI Visibility package for me? I could do with some testing on a larger site to see how it performs https://marketplace.umbraco.com/package/umbraco.community.aivisibility Let me know if you do try it and send me any feedback! It basically does a few things: - /llms.txt + /llms-full.txt manifests auto-generated per the llmstxt.org spec - Every published page reachable as Markdown via Accept: text/markdown or by appending .md to the URL - AI Traffic dashboard in the Backoffice — every AI / human / crawler hit by classification, with date filtering - Settings dashboard for site name, summary, per-doctype + per-page exclusion (no appsettings.json edits needed) - Robots audit Health Check that warns when robots.txt blocks AI crawlers (audits only, never modifies your robots.txt) - 6 DI extension points to override extraction, request logging, robots auditing, settings resolution, UA classification — without forking - Zero configuration on a typical Umbraco v17+ site. Apache 2.0 licensed.

@adam I would be interested in testing it on codeshare. When does it write to the database? I am trying to work out if it happens based on page views or in other ways, or not at all?

Currently writes to aiVisibilityRequestLog happen only on successful 2xx responses to the package's three Markdown surfaces: - /llms.txt - /llms-full.txt - Per-page Markdown (either .md direct, or Accept: text/markdown content negotiation on the canonical URL) So it's only AI-crawler / Markdown-route traffic. NOT regular HTML page views. Standard /about/ browsing by humans doesn't touch the log. Other relevant bits: - Skipped on 304 / 404 / 500 (cache revalidation + miss paths don't double-count) - Async fire-and-forget — bounded channel drained in batches by a hosted service, doesn't block response time - 90-day retention by default; background job purges older rows - Kill switch via AiVisibility:RequestLog:Enabled = false - IRequestLog is a public extension point if you want to redirect writes to App Insights, Serilog, or a different sink

I guess I took the assumption that traffic to the .md files will be failry light. humans won't visit them only AI Spiders really, so DB write should be fairly low

I am having a massive issue with AI traffic at the moment. 25GB of bandwidth per day. Trying to block it. It would be most of my traffic. So I think anything that writes to the db when they visit would kill my site

oh really! well I guess this might help if they are redirectd to the md files? the HTML traffic isn't written? Claude says 🙂 Realistic volume: - Humans never hit .md URLs (they get HTML on /about/, which doesn't write anything). Only AI crawlers + audit tools hit the Markdown surfaces. - AI crawler hit rates are roughly the same order as traditional search bots (Googlebot, Bingbot) — hundreds-to-low-thousands/day for a content site, not orders of magnitude more. - 304 cache revalidations skip writing entirely. Cloudflare-fronted sites have most of their AI traffic filtered before it reaches the writer. Defences in the package: - Bounded channel with DropOldest — default 1024 capacity, configurable to 65536. Under pressure, oldest entries drop; never blocks the request thread or blows memory. - Batched writes via NPoco InsertBulk: 50 rows per batch OR every 1s, whichever fires first → roughly 1 INSERT/s per node under sustained load, not 1 per request. - 90-day retention purge by default; configurable. - Each row is small (~150-300 bytes — path + content key + culture + UA class + referrer host). For high-traffic adopters, three escape hatches: - Tune the knobs: AiVisibility:RequestLog:{QueueCapacity, BatchSize, MaxBatchIntervalSeconds} + LogRetention:DurationDays - Kill the writer entirely: AiVisibility:RequestLog:Enabled = false - Replace IRequestLog (public DI extension point) — point at App Insights with built-in sampling, wrap with a sampling decorator (e.g. log 1 in 10), or redirect to your own store For codeshare-scale: defaults handle it without thinking. For a major news org pushing 500k AI hits/day: tune knobs or swap IRequestLog. No adopter is forced into the default shape.

Very fun and neat project! Thanks for sharing 👍

@adam I'm definitely trying this! Couple of questions: * Could I block agents if they are being bad actors (way too many requests for md or llm files)? * Can I download a CSV of a the information? Ideally a date range (probably a month of data)?

Sorry! Forgot to reply. - No currently you can't block agents, i can look at adding that to the roadmap maybe - no, you can't download a CSV -good idea, i will add that too! A new release has been pushed out today if you want to look at that.

Quick update for anyone running, or wants to run, my package Umbraco.Community.AiVisibility. v1.1 is out and the big change is a dual-strategy renderer that now handles real agency-built sites properly, not just clean starter-kit installs. The new Auto default tries the fast in-process render first and falls back to an HTTP loopback render automatically, so route hijacking, custom controllers and custom view models all work without any config changes from adopters. Verified on a production-grade Umbraco install with all the usual real-world complexity. Also fixed a latent v1.0 bug where pages locked to logged-in members were leaking their login-redirect HTML to AI crawlers as if it were normal Markdown content. Now correctly excluded from the .md route, /llms.txt and /llms-full.txt. If you're already running it, you'll pick up the changes on your next package update. Feedback and issues very welcome. Would love some feedback if you have installed it or feel free to raise any issues on GitHub or feature reequests!

Building support for the library in Advanced Permissions v18 😄 https://cdn.discordapp.com/attachments/905057811682193429/1516392682790981683/image.png?ex=6a327a3c&is=6a3128bc&hm=601531d848e4f29f7da2f4482eaf40ae43bb981210b74f3b31bd9a8fd1ae7eda&

Released a great update for the Discipline package today - everything is now configurable via the back office. Details here: https://www.dot-see.com/blog/mega-update-dotseediscipline-ui-based-configuration/

@UMB.FYI tip

Thanks for the tip

Here's a video I posted on LinkedIn showing how to set up Claude Design with the UUI library that I demoed in the Codegardent talk. https://www.linkedin.com/feed/update/urn:li:ugcPost:7478382465955500033/

Oooo thanks Phil, I was hoping you would show how to do this as it looked really impressive in your CG talk

Here you are @Warren Buckley https://cdn.discordapp.com/attachments/905057811682193429/1522291173308370944/Claude_Design_Setup.mp4?ex=6a47efa2&is=6a469e22&hm=e522d6cab70d1d2225a119738a1a4a014b6fe7d5b51c107b610b04d517ba03d9&

Yesterday I launched the updated version of my Umbraco blog https://cornehoskam.com 🙌 This includes an upgrade from Umbraco 13 to 17, but most notably it launched with a complete redesign made possible by Claude Design! (both of which I'll be writing about soon!) All in all it now has a more consistent color palette across the board while staying true to the old 'branding', more emphasis on the blog post & content itself with fewer distractions, improved SEO & A11Y, and a smoother experience overall, lowering the threshold for me to write new blogs in the future 🚀 https://cdn.discordapp.com/attachments/905057811682193429/1524674797152243902/HNKchPs37g.png?ex=6a509b8e&is=6a4f4a0e&hm=3f5e82e758bf1c0fd8906bd1b360df441755fcbf4328c1ed3f7c2e688caa177e&

One interesting post from me that you can share with all those potential clients that have been on the SharePoint train for the last decade or so: https://www.dot-see.com/blog/why-umbraco-cms-is-a-strong-choice-for-modern-intranet-portals/

In other news, I released this package - easy generation of responsive images with one-setting switch to Cloudflare URLs generation, lazy loading, blurs, loading placeholders, rules, the works. We've been using it for a long time now as part of our builds, but it was about time it became a package. https://www.nuget.org/packages/dotsee.responsiveimages#readme-body-tab

Finally, a fix that took 10(!) versions to add is now on DotSee.Discipline: NodeRestrict working with top-level nodes. https://www.nuget.org/packages/DotSee.Discipline

New blog post on Umbraco website implementation costs https://www.dot-see.com/blog/how-much-does-an-umbraco-website-implementation-cost/

(It looks like I've been monopolizing this channel lately, sorry 🙂 ) We created an interesting, although still beta, Umbraco Site Checker at https://www.dot-see.com/check-your-umbraco-sites-health/ - not intrusive - checks your public page only, and gives you useful insights from what it can find about your site - only if it is an Umbraco site. It cannot access your back office or provide detailed information on what's going on with your add-ons, but you'll be surprised how many things it can find just from the front-end. Yes, we store your query (emal + url), but only to be able to check against repeated / automated requests as well as send you the report. We will never use it to send you additional emails (unless you check the optional checkbox). Give it a spin - and let me know how it goes. Thanks!

OK, this is the first time I've done anything like this so I'm a bit nervous and hopefully I'm putting this in the right place(!) - with a bit* of help from Claude I've created a resource and service booking package for Umbraco: *rather a lot - I've been a developer for ~30 years and working with Umbraco since v7 but most of what I've written has been for clients and therefore not public, and since I have clients who keep me busy Claude has been a bit of a godsend: https://github.com/Chris-N2/UBookIt It's usable now and installable on Umbraco v17 sites - I'm just writing the roadmap for v17.1.0 that will add things like editing bookings (rather than just deleting them) in the back office and a proper admin settings screen. Next will be a DevExpress based package that will replace the UI - it has theming built in so you can create your own 🙂

Looks nice! It would help to add a few screenshots to your readme.md. Also, add it to the marketplace 🙂

Thanks - those are both on the list, there's just a few more functional bits to get finished first 🙂

Last weekend I tried to use Claude for making an Umbraco plugin with the new backoffice and to my surprise it works pretty well! I was able to make 3 plugins that I've always wanted to make based on my experience with Umbraco and working with customers over the last 10 years. I'm sure any of you can find at least 1 plugin that will enhance your Umbraco experience. TrueCopy - https://marketplace.umbraco.com/package/our.umbraco.truecopy Ever needed to copy a whole site within Umbraco? What's the first thing you do after the pages are copied? You go into the new pages and fix all the links because they still point to the original pages. This plugin will go through all the copied pages and point the links to the new pages. ErrorDashboard - https://marketplace.umbraco.com/package/our.umbraco.errordashboard Browsers can send a report to your website in case they encounter a 404, 500, SSL or connection issue. This plugin collects those reports and sends an email in case there is a sudden spike of errors. (This only works with Chrome-based browsers for now. Firefox might support it in the future, it's behind a flag) ContentDashboard - https://marketplace.umbraco.com/package/our.umbraco.contentdashboard This is for when you have a whole team of content editors working on the site. You can assign/transfer ownership of pages to users and see which pages are scheduled to be (un)published. I made the plugins for Umbraco 18 instead of 17 LTS since they are all new and not battle tested yet. If you want to try out a plugin without installing it on your own site you can easily clone the repository and run the included sample website.

All three are great - but it would be amazing if they were also available for v17, since v18 is not picked for production by everyone yet. I'll be sure to give them a go anyway though.

Do I have some good news for you: they are all available for Umbraco 17 (17.7.0+) as of today!

Feel like showing off some of the ridiculous thing I built for my blog-post https://www.kjeldsen.dev/blog/the-umbraco-mcp-benchmark-or-the-little-haiku-that-could/ - little things that aren't big enough to present other places 😄 First off I have created a generated graphics media type in Umbraco which is hooked into Umbraco AI I write a prompt - can give it an example image, aspect ratio, choose whether it should animate and give it animate directions and then you end up with something like the little video clip I've attached 🙂 Charts in the post are made with the same principle, where I have a "data visio block" in which I can give it a dataset, litterally just has to be json, direction as to what I am trying to show, a chart type and off it goes and creates a chart that is now a movable block in my backoffice https://cdn.discordapp.com/attachments/905057811682193429/1556926408687751218/claude_hill.mp4?backend=b2&ex=6ac5f036&is=6ac49eb6&hm=10ebdc44e901159795268efe0db45d345b7617792e128853b05ea30bb5c93843& https://cdn.discordapp.com/attachments/905057811682193429/1556926409597788191/image.png?backend=b2&ex=6ac5f036&is=6ac49eb6&hm=5fde23e432d22695c60cc2e33c4ab774588317d31ecf0d4b2d530bf8a0e98e63& https://cdn.discordapp.com/attachments/905057811682193429/1556926410877173781/image.png?backend=b2&ex=6ac5f036&is=6ac49eb6&hm=59380c3458bfbd3919e300fa49725a0ea22391a9c3ed85515cea684a0dfd652e& https://cdn.discordapp.com/attachments/905057811682193429/1556926411854323732/image.png?backend=b2&ex=6ac5f036&is=6ac49eb6&hm=135988a7df81aeb78c14d509ba2b184459603ec86db12660e48580edcfdbdc3b&