Alessio Vertemati

@rauchg Content-md contentmd.org website layout and the command line. Used Claude Code with Opus 4.6

Show me the thing you’ve built with AI you’re most proud of. Reply with a working product URL and what model / agent you primarily used.

The module check the accept header and serve the pre-computed markdown file if found. I'm using it two power contentmd.org

Serving markdown to agents from static websites, easy as @caddyserver can do that, just adding options on top of a solid base github.com/avvertix/caddy…

@_philschmid Vibes is the starting point for deciding what to evaluate first and whether to evaluate it. Unless we speak about unit-tests

Vibes > Evals or Evals > Vibes?

@mattshumer_ Agreed! While working with agents, Markdown improves readability and enables fast interaction. HTML brings style, which we usually focus on at the expense of the text we were working on. Few thoughts on what I include in Markdown contentmd.org

Everyone switching from Markdown to HTML is missing the nuance. The optimal approach (most of the time): If it's for a human to read, yeah, use HTML. BUT If an agent is consuming it, use Markdown!

I think it's a matter of reading versus presenting. Markdown is simple enough to be read, copied and pasted across different editing workflows. HTML is the final artefact; once you're in HTML, you start thinking about colours and the text goes into the background.

@karpathy HTML can be tricky when you want your output to be copyable elsewhere. I've noticed that Claude prefers to write HTML, perhaps due to a prompt configuration, when chatting. Personally, I prefer to write in Markdown first and convert later.

This works really well btw, at the end of your query ask your LLM to "structure your response as HTML", then view the generated file in your browser. I've also had some success asking the LLM to present its output as slideshows, etc. More generally, imo audio is the human-preferred input to AIs but vision (images/animations/video) is the preferred output from them. Around a ~third of our brains are a massively parallel processor dedicated to vision, it is the 10-lane superhighway of information into brain. As AI improves, I think we'll see a progression that takes advantage: 1) raw text (hard/effortful to read) 2) markdown (bold, italic, headings, tables, a bit easier on the eyes) <-- current default 3) HTML (still procedural with underlying code, but a lot more flexibility on the graphics, layout, even interactivity) <-- early but forming new good default ...4,5,6,... n) interactive neural videos/simulations Imo the extrapolation (though the technology doesn't exist just yet) ends in some kind of interactive videos generated directly by a diffusion neural net. Many open questions as to how exact/procedural "Software 1.0" artifacts (e.g. interactive simulations) may be woven together with neural artifacts (diffusion grids), but generally something in the direction of the recently viral x.com/zan2434/status… There are also improvements necessary and pending at the input. Audio nor text nor video alone are not enough, e.g. I feel a need to point/gesture to things on the screen, similar to all the things you would do with a person physically next to you and your computer screen. TLDR The input/output mind meld between humans and AIs is ongoing and there is a lot of work to do and significant progress to be made, way before jumping all the way into neuralink-esque BCIs and all that. For what's worth exploring at the current stage, hot tip try ask for HTML.

@_philschmid Agree, Markdown plus components helps control token usage. Rendering is the last part, most of the time you need readability first. With contentmd.org I'm proposing somehow a bridge

Feels like MDX might be the right middle ground in terms of expressioness, token control and ease of use. Yes rendering is slighty more complicated.

@addyosmani I think html is best as presentation layers at the end of the process, while working with agents markdown is quicker to read and edit. That's how I created contentmd.org started in markdown than iterated in html for the design

Move from Markdown to HTML and give AI a richer canvas to communicate output humans will actually read. Think plans, specs even throwaway editors. Like this idea which keeps more humans in the loop.

@ashleyhindle @every I also considered agent readiness and made a proposal for what the returned markdown should look like contentmd.org

Liked this thought from Marc about startups being agent-ready 🔑 @every are a good example of what this looks like in practice, will be stealing some of their lessons 🤫 Here's a very quick look 👀

My first #Caddy module. Static websites can now serve markdown to agents github.com/avvertix/caddy…

@JohnONolan @Ghost Time flies, I was actually one of the 5,000+. The campaign video was ahead of its time.

On this day, 13y ago, @ghost started as an idea on kickstarter 5,000+ people backed the project to help bring it to life and support the future of media and journalism Today, Ghost has generated over $150M for independent publishers all over the world kickstarter.com/projects/johno…

@retlehs @Cloudflare Cool! Didn't know about the 406 status! Adding it to my Caddy module github.com/avvertix/caddy…

The new @Cloudflare site is a great baseline — it checks whether you return Markdown ✨ acceptmarkdown.com checks whether you return it correctly (Vary, q-values, 406, Link rel=alternate), tracks which AI agents actually adopt the standard, and includes integration guides

@pushpak1300 Very interesting! Did the AI read the tests during development or had a look at the test folder during the implementation?

Everyone asks “which AI model should I use?” We tested them with laravel and boost so you don’t have to.

Something is coming 🦞 Any interaction with this tweet will be considered for free credits.

@matt_elevenlabs What an amazing journey! 11B

We just raised $500M at an 11B valuation 🎉 To celebrate, we’re giving away 1,000 free credits so you can test our platform. For the next 6 hours, comment “11B” below and we’ll DM you the credits (must follow) 👇

Want to install agent skills in your project? Using Python? Got you covered, vibe ported the npx skills to uvx skillsmd, same commands github.com/avvertix/skill…

@huggingface @Gradio @AnthropicAI That's my take for photographers huggingface.co/spaces/MCP-1st…

Time to see if I can build a MCP server. Cool that there is the MCP's 1st birthday hackathon with @huggingface @Gradio and @AnthropicAI huggingface.co/MCP-1st-Birthd…