I Can't Believe Cloudflare Actually Shipped This

I Can't Believe Cloudflare Actually Shipped This ![rw-book-cover](https://img.youtube.com/vi/X8yVdd7lGmw/maxresdefault.jpg) 🔗 **Source:** [I Can't Believe Cloudflare Actually Shipped This](https://www.youtube.com/watch?v=X8yVdd7lGmw) > [!note]+ Document Notes > tags: #wordpress #cloudflare #cms #astro #serverless > author: Ahmed > source: YouTube > reference: https://www.youtube.com/watch?v=X8yVdd7lGmw > date created: 2024-12-19 14:30 > > # 🏷️ Title: I can't believe Cloudflare actually shipped this >  ## 🧰 Action Inbox - [H1] Test Mdash playground thoroughly — security and content management concerns #cms - [H2] Compare serverless limits vs traditional hosting — cost analysis needed #serverless - [H3] Evaluate WordPress plugin ecosystem migration — developer adoption strategy #wordpress ## 🧭 Table of Contents - [[#📝 Summary]] - [[#📖 Main Content]] - [[#🗒️ Notes]] - [[#📚 References]] - [[#🗣️ Paraphrase]] - [[#📰 Newsletter]] - [[#✅ Action Items]] - [[#🧪 Hypotheses]] - [[#📊 Methodology]] - [[#❓ Questions]] - [[#🔮 Future Research]] - [[#💡 Implications]] - [[#🧠 Learning Styles Assessment]] - [[#📊 Accessibility Assessment]] - [[#🌍 DEIB Assessment]] - [[#✅ Logical Validity]] - [[#🛡️ DARVO Analysis]] - [[#🎯 Conclusion]] ### 🔗 References - [Mdash CMS](https://blog.cloudflare.com/mdash) - [WordPress Core](https://wordpress.org) ## 📝 Summary - Cloudflare launched Mdash, a TypeScript-based CMS claiming to be a "WordPress spiritual successor" - The author, an ex-WordPress core developer, finds significant issues with the playground and positioning - Key concerns include content loss bugs, poor UI copying WordPress design, and unfair security criticisms - While serverless architecture has benefits, vendor lock-in and platform limits create new challenges - The future may favor custom-built solutions using AI agents rather than pre-built CMS platforms ## 📖 Main Content Ahmed, an experienced WordPress core developer and founder of Command Code, provides a critical analysis of Cloudflare's new Mdash CMS. Despite Cloudflare's track record of launching legitimate projects on April 1st, this WordPress competitor raises serious concerns about both execution and positioning. The core tension lies between WordPress's monolithic, GPL-licensed approach serving 40% of the web versus Mdash's serverless, TypeScript-based architecture running on Cloudflare Workers. While WordPress succeeded through easy installation, extensive plugin ecosystem, and backward compatibility, Mdash attempts to solve perceived security and scaling issues through isolated dynamic workers. However, the implementation reveals significant problems. The playground demonstration shows content management failures, UI accessibility issues, and a poorly executed WordPress dashboard clone. More concerning is Cloudflare's messaging that positions WordPress plugins as a "security crisis" - a claim the author finds distasteful and technically questionable. The serverless approach introduces new complexities including vendor lock-in, platform limits (500MB database for free tier), and unpredictable costs based on request volume rather than fixed hosting fees. These limitations may prove more restrictive than the traditional LAMP stack problems Mdash aims to solve. ## 🗒️ Notes #### ⏱️ [00:00](https://youtu.be/X8yVdd7lGmw?t=0) — [H1] Cloudflare launches WordPress competitor - Key idea: Mdash CMS announced as "spiritual successor" to WordPress with TypeScript and serverless architecture - Evidence: "TypeScript which is serverless MIT license and built for the agent era" #### ⏱️ [02:30](https://youtu.be/X8yVdd7lGmw?t=150) — [H2] Author's WordPress background and concerns - Key idea: Decade-long WordPress core developer questions the competitor's approach and messaging - Evidence: "I spent more than a decade building the core software of WordPress" #### ⏱️ [05:20](https://youtu.be/X8yVdd7lGmw?t=320) — [H1] WordPress's historical success factors - Key idea: WordPress won through easy installation, plugin ecosystem, and hosting infrastructure - Evidence: "WordPress had this famous 5-minute install and this was one of the earliest wins" #### ⏱️ [08:45](https://youtu.be/X8yVdd7lGmw?t=525) — [H2] Developer exodus from WordPress ecosystem - Key idea: Modern developers moved to React/TypeScript while WordPress maintained backward compatibility - Evidence: "developers kind of left the WordPress ecosystem when they found it to be very very sticky" #### ⏱️ [12:15](https://youtu.be/X8yVdd7lGmw?t=735) — [H1] Playground demonstration reveals serious issues - Key idea: Content management failures and poor user experience in Mdash demo - Evidence: "You're building a content management system, CMS. You cannot lose my content" #### ⏱️ [16:30](https://youtu.be/X8yVdd7lGmw?t=990) — [H3] UI criticism of WordPress dashboard clone - Key idea: Poor design execution that looks like "vibe coded" WordPress admin - Evidence: "This looks like a poorly vibe coded WordPress dashboard. This does not look good" #### ⏱️ [20:10](https://youtu.be/X8yVdd7lGmw?t=1210) — [H2] Security claims analysis - Key idea: WordPress plugin security criticism is distasteful and technically questionable - Evidence: "96% of WordPress issues are happening due to plugins... This is like saying 96% of accidents happen on cross-sections" #### ⏱️ [25:40](https://youtu.be/X8yVdd7lGmw?t=1540) — [H1] Serverless architecture trade-offs - Key idea: Worker-based isolation provides security but introduces platform limits and vendor lock-in - Evidence: "500 workers in your Cloudflare account... Everything serverless has weird limits to it" #### ⏱️ [30:20](https://youtu.be/X8yVdd7lGmw?t=1820) — [H2] Cost comparison concerns - Key idea: Serverless pricing model may be more expensive than traditional hosting - Evidence: "the amount of automations that I run would render more than $25 bill on D1 alone" #### ⏱️ [35:15](https://youtu.be/X8yVdd7lGmw?t=2115) — [H3] WordPress democratization spirit - Key idea: WordPress's mission differs fundamentally from Cloudflare's approach - Evidence: "WordPress exists to democratize publishing... Anybody should be able to do it" #### ⏱️ [40:30](https://youtu.be/X8yVdd7lGmw?t=2430) — [H1] Future of CMS development - Key idea: AI agents enable custom solutions, reducing need for pre-built platforms - Evidence: "I can just talk to any coding agent and I can say build me this this and that" #### ⏱️ [43:45](https://youtu.be/X8yVdd7lGmw?t=2625) — [H2] Author's own CMS project - Key idea: Two-day TypeScript WordPress-compatible API build demonstrates AI-enabled development - Evidence: "I was able to build this within two days... mimics the entire WordPress API" ## 📚 References - [Cloudflare Mdash Blog Post](https://blog.cloudflare.com) - [WordPress.org](https://wordpress.org) - [Astro Framework](https://astro.build) - [Command Code](https://commandcode.com) ## 🗣️ Paraphrase - Cloudflare's Mdash CMS attempts to modernize WordPress with TypeScript and serverless architecture, but fails in execution with content management bugs and poor UI decisions that copy WordPress without understanding user needs - The positioning attacks WordPress plugin security unfairly while introducing new complexities like vendor lock-in, platform limits, and unpredictable serverless costs that may be worse than traditional hosting problems - The future likely favors AI-enabled custom development over pre-built CMS platforms, as developers can now build exactly what they need rather than adapting existing boilerplate solutions ## 📰 Newsletter **Cloudflare's WordPress Alternative: Promise vs Reality** Cloudflare launched Mdash, a TypeScript CMS claiming to be WordPress's "spiritual successor." While the serverless architecture and modern stack sound appealing, early testing reveals significant content management issues and questionable positioning against WordPress. Key concerns for WordPress developers: - Playground demos show content loss and UI accessibility problems - Unfair criticism of WordPress plugin security ignores broader software reality - Vendor lock-in and serverless limits may create new constraints - Missing plugin ecosystem that made WordPress successful The real insight: AI agents now enable custom CMS development in days rather than months, potentially reducing demand for pre-built platforms entirely. **What this means for WordPress:** The ecosystem remains strong, but developers increasingly build custom solutions. Focus on what WordPress does best - democratizing publishing for non-technical users. Resources: [Learn WordPress](https://learn.wordpress.org/), [Developer Docs](https://developer.wordpress.org/), [WP-CLI](https://wp-cli.org/) ## ✅ Action Items #### WordPress - Core: Evaluate modern developer experience gaps, TypeScript integration opportunities - Core Performance: Benchmark against serverless alternatives, document scaling capabilities - Community: Address developer retention through modern tooling and workflows - Documentation: Create serverless deployment guides, headless CMS patterns - Plugins: Improve security review processes, developer guidelines for secure coding - Meta: Study competitive positioning, emphasize democratization mission #### DevRel - Developer advocacy: Counter unfair security narratives with data-driven comparisons - Education: Create workshops on WordPress vs serverless trade-offs - Developer experience: Modernize toolchain, improve TypeScript support - Community management: Monitor discussions around WordPress alternatives ## 🧪 Hypotheses - Pre-built CMS platforms will lose market share to AI-generated custom solutions within 2-3 years - Serverless CMS costs will exceed traditional hosting for content-heavy sites due to request-based pricing - WordPress will maintain dominance in non-technical user segment despite developer migration - Vendor lock-in concerns will limit enterprise adoption of platform-specific CMS solutions ### 📊 Methodology - Cost analysis comparing WordPress hosting vs Cloudflare Workers pricing across usage patterns - Survey WordPress developers on pain points vs serverless migration barriers - Performance benchmarking between monolithic and serverless CMS architectures ### ❓ Questions - How will WordPress adapt its developer experience to compete with modern TypeScript alternatives? - What are the true total costs of serverless CMS deployment including data egress and compute charges? - Can isolated worker architecture actually prevent the security issues it claims to solve? ### 🔮 Future Research - Long-term viability of AI-generated vs pre-built CMS solutions - Enterprise adoption patterns for serverless vs traditional CMS architectures - Developer productivity comparison between WordPress ecosystem and modern alternatives ### 💡 Implications - Practice: WordPress developers should evaluate modern tooling gaps and consider TypeScript integration - Policy: CMS vendor claims about security should be evaluated critically with supporting data - Security: Isolation models require thorough testing beyond theoretical guarantees - Education: Developers need training on serverless cost models and platform limitations - Funding: Investment in WordPress modernization may be necessary to retain developer mindshare ### 🧠 Learning Styles Assessment Reading level estimate: 10th grade #### 👀 Visual learners - Present: Architecture diagrams comparing monolithic vs serverless approaches - Improve: Cost comparison charts, security model visualizations, platform limit infographics #### 🔊 Aural learners - Present: Video format with detailed verbal explanations and live demonstrations - Improve: Audio-only version for podcast consumption, shorter summary segments #### 🗣️ Verbal learners - Present: Technical terms explained with analogies, clear definitions of CMS concepts - Improve: Glossary of WordPress vs serverless terminology, comparison frameworks #### 🧩 Kinesthetic learners - Present: Live playground demonstration, hands-on testing of both platforms - Improve: Step-by-step migration guides, interactive cost calculators #### 🧠 Logical learners - Present: Structured comparison framework, cost-benefit analysis approach - Improve: Decision trees for CMS selection, quantitative performance metrics #### 🧑‍🤝‍🧑 Social learners - Present: Community discussion context, developer ecosystem perspectives - Improve: Forum integration, peer review opportunities for CMS evaluations #### 🧘‍♀️ Solitary learners - Present: Detailed individual analysis, personal experience narratives - Improve: Self-assessment checklists, reflective questions on CMS needs #### 🌿 Naturalistic learners - Present: Evolution of CMS ecosystem, natural progression of technology - Improve: Lifecycle thinking about platform sustainability, ecosystem health metrics ### 📊 Accessibility Assessment #### Screen reader semantics - Present: Clear heading structure, descriptive video content - Improve: Transcript availability, better alt text for technical diagrams #### Keyboard navigation - Present: Video controls accessible, logical content flow - Improve: Skip links for different topics, keyboard shortcuts for sections #### Contrast and readability - Present: Good text contrast in video player - Improve: High contrast mode support, larger text options #### Text resizing and reflow - Present: Responsive video player, scalable interface - Improve: Text-only version of key points, better mobile experience #### Media and captions - Present: Full video content with audio - Improve: Accurate captions, downloadable transcript, chapter markers #### Forms and interactions - Present: Basic video interaction controls - Improve: Accessible feedback forms, better error handling ### 🌍 DEIB Assessment #### Cultural breadth - Present: Global perspective on WordPress usage, international hosting examples - Improve: Multi-language CMS considerations, regional hosting preferences #### Inclusive language - Present: Technical terms explained clearly, avoiding excessive jargon - Improve: Plain language summaries, cultural context for technology adoption #### Neurodiversity support - Present: Structured presentation with clear sections - Improve: Content warnings for rapid UI changes, consistent navigation patterns #### Representation - Present: Developer perspective from experienced practitioner - Improve: Multiple viewpoints from different user types, diverse use cases #### Psychological safety - Present: Respectful criticism focused on technical issues - Improve: Balanced perspective acknowledging both platforms' strengths ### ✅ Logical Validity #### Strong reasoning - Deductive strength: Technical analysis follows from platform specifications and testing results - Inductive strength: Conclusions drawn from extensive WordPress development experience - Causal reasoning: Clear connections between design decisions and user experience outcomes - Analogical strength: Effective comparisons between CMS approaches and broader software trends #### Fallacies - Ad hominem: Minimal personal attacks, criticism focused on technical merits - False dichotomy: Avoids presenting WordPress vs Mdash as only options, acknowledges spectrum #### Tone of voice - Overall tone: Critical but constructive, technically informed with practical concerns ### 🛡️ DARVO Analysis - Deny: No significant denials of legitimate concerns - Attack: Criticism focuses on technical execution rather than attacking Cloudflare's intentions - Reverse victim/offender: No role reversals identified - Emotional cues and rhetoric: Passionate but technically grounded criticism ## 🎯 Conclusion Cloudflare's Mdash represents an interesting technical approach to CMS architecture, but falls short as a WordPress successor due to execution issues and positioning problems. The real insight is that AI-enabled development may reduce the need for pre-built CMS platforms entirely, as developers can now create custom solutions tailored to specific needs. WordPress's strength lies in democratizing publishing for non-technical users, while the developer ecosystem increasingly moves toward bespoke solutions. ## 🧱 Refactor Appendix ### 📚 Literature Split Pad Purpose: Stage CEE units that become Literature notes. Split level: H3 ### [H1] WordPress succeeded through ecosystem effects not technical superiority > [!quote] I can't believe Cloudflare actually shipped this by Ahmed via YouTube. Published 2024-12-19. Accessed 2024-12-19. > "WordPress was successful because one of the biggest reasons was easy to install, plugin ecosystem, and hosting infrastructure." Historical analysis of WordPress adoption factors that new platforms must match. [view highlight](https://youtu.be/X8yVdd7lGmw?t=320#H1) #xref #articleseed ID:rw_wp_success ### [H2] Serverless CMS introduces new constraints while solving old ones > [!quote] I can't believe Cloudflare actually shipped this > "Everything serverless has weird --- ## Full Transcript [00:00] Cloudflare seemed to have launched a [00:02] WordPress competitor and they chose [00:04] April 1st to do it. Is it actually a [00:07] joke or is it something that we should [00:10] actually be calling a spiritual [00:12] successor for WordPress? That is what [00:14] I'm going to discuss today in this [00:16] video. Hey, I Ahmed and I'm an open [00:18] source developer and founder of Command [00:20] Code, a coding agent with taste. So, I [00:22] have deep interest in discussing this [00:24] topic because I spent more than a decade [00:26] building the core software of WordPress. [00:27] I have a lot of thoughts and opinions [00:29] and we are going to unpack all of what [00:32] is written in the introduction to M dash [00:35] uh you know CMS post from cloudfare in [00:37] full disclosure I'm founder of command [00:39] code and if you go to our about page you [00:41] will actually see that we have great [00:43] investors but one of those great [00:44] investors is actually Dane so we love [00:47] what Cloudflare is doing you know this [00:49] goes to say that Cloudflare also [00:50] supports you know what we are building [00:52] it on a face cloudflare has this history [00:55] of launching projects instead of April [00:57] fool jokes on 1st of April every year. I [01:00] would know because I was one of the [01:02] earliest users of Cloudflare. So when I [01:04] built Langbase in command code, we chose [01:06] Cloudflare as a host to bunch of [01:08] infrastructure. So when Dean went ahead [01:10] and announced this announcing mdash, the [01:13] WordPress virtual successor built for [01:15] modern web with TypeScript which is [01:17] serverless MIT license uh and built for [01:21] the agent era. I had a lot of thoughts, [01:23] right? I feel like you cannot just say [01:25] these things so lightly nowadays, right? [01:27] And this is coming right off the bat [01:28] after Cloudflare actually vibe coded a [01:32] Nex.js in a competitor called Vinex [01:35] within a week and was kind of called out [01:38] for it. This one however is a little bit [01:40] different. So this is a tweet from Dean [01:41] and I instantly replied like I don't [01:43] know if this comes close to being a [01:45] WordPress successor. I mean the idea is [01:48] no, don't get me wrong, but there are [01:50] many things that are wrong about this. I [01:52] feel like the WordPress like look of [01:54] their dashboard is throwing me off and [01:57] the entire uh you know stab at WordPress [01:59] for security and plugins. I don't know [02:01] about that. So let's dive a little bit [02:03] deeper into this and discuss WordPress [02:06] versus mdash. One of the first things [02:08] I'm going to start with is if you take a [02:10] look at this WordPress is a tool the [02:12] open-source content management system [02:14] content has this innate ability. It [02:17] requires to be managed. The moment you [02:20] start working with a bunch of team [02:21] members and you are writing a lot of [02:23] content, you would come to this [02:25] realization as well. I started very [02:28] early on WordPress. I started uh writing [02:30] blogs maybe as early as 2001 and most of [02:32] my blogs are actually built with [02:34] WordPress. When I was in my early 20s, [02:36] WordPress was the cool kid in town, [02:38] right? There was Jumla, there was Drupal [02:40] and there was a bunch of other mess like [02:42] the blogger platform from Google, but [02:45] WordPress was it at that moment. Like [02:47] you know how typescript and react are so [02:49] cool nowadays or agentic engineering is [02:52] so cool nowadays that is what we felt [02:54] like working on WordPress. So now a lot [02:57] has changed. WordPress is 24 years old. [02:59] WordPress is right now still 40% of the [03:02] internet. If you go to this W3 text [03:05] website you will see that 60% of all the [03:08] CMS uh in probably top 10 million [03:11] websites is still WordPress. And that is [03:14] almost 42% like almost half of the web [03:17] is WordPress. I think WordPress is [03:20] successful because one of the biggest [03:22] reasons was easy to install. WordPress [03:24] had this famous 5minut install and this [03:28] was one of the earliest wins. And then [03:30] the second thing that kind of jumped on [03:32] me was plug-in ecosystem. It was open [03:34] source or GPL friendly. It was super [03:37] easy to install. So when I started [03:39] working with WordPress, I found it very [03:41] easy that I can easily test a bunch of [03:43] different plugins and I don't have to go [03:45] through the entire FTP route or whatnot. [03:48] And then came the entire extensible [03:51] infrastructure around this like the [03:53] first thing was custom post types, CPTs, [03:56] right? Where you know by default [03:58] WordPress was actually just built for [04:00] writing blog posts or you know creating [04:02] pages. But now with custom post types [04:04] you could create different things. You [04:06] could sell things. You could create a [04:07] books custom post type. You could create [04:10] you know a software custom post type [04:11] whatnot. And WordPress adoption actually [04:14] completely uh you know spread like a [04:17] fire because of that thing. And then the [04:19] rest of it was maybe around 2011 or 10 [04:23] when we got the rest API in WordPress [04:25] and we started using it as a headless [04:28] CMS of sorts. And then around the time [04:30] when Gutenberg happened I was very [04:32] deeply involved in the Gutenberg [04:34] project. uh WordPress actually started [04:36] adopting earlier version of faxjs which [04:39] became reactjs. So and then WordPress [04:43] actually uh got Facebook to open source [04:47] react. I remember creating the issue on [04:49] GitHub GitHub going down because of this [04:51] discussion around that issue and how [04:53] WordPress kind of influence uh a huge [04:57] community and a huge uh tech form like [04:59] Facebook at the time to open source both [05:02] GraphQL and React, right? So WordPress [05:04] has this amazing ecosystem of developers [05:07] uh plug-in and you can also say themes [05:10] that made it really really fun to use. [05:12] But around five or six years ago [05:14] something started to break. You know the [05:16] ecosystem started to go in a different [05:18] direction and developers started to lose [05:22] interest in WordPress. Developers [05:23] started to build with React and [05:26] TypeScript. I feel like developers kind [05:28] of left the WordPress ecosystem when [05:30] they found it to be very very sticky. um [05:32] and the backward compatibility was a big [05:34] issue or whatnot. And since then while [05:37] WordPress has continued to grow in the [05:39] nontechnical community, the super [05:40] admins, the people who are not actually [05:43] uh ready to pay for custom software, who [05:45] depend on a lot of themes and uh you [05:47] know uh custom plugins all day long, but [05:49] the modern cutting stack has kind of [05:52] moved away from the CMS, right? I think [05:54] that is sort of what uh Cloudflare is [05:57] trying to do. So Cloudflare has [05:59] introduced MDS. Cloudflare just recently [06:01] acquired Astro. Astro is a framework [06:03] that is built around content-driven [06:05] websites, basically blogs, right? If [06:07] you're writing a lot of blogs or [06:08] building a lot of fast marketing [06:10] websites, Fred and team did an amazing [06:12] job in building Astro and Astro can work [06:14] with anything. It can work with Vue, uh, [06:17] ReactJS, I think Swelt as well, and can [06:19] even work without JavaScript, right? So, [06:21] it's a static site generation framework [06:24] of sorts. So right after acquiring Astro [06:27] I think Cloudware is actually trying to [06:29] capture the CMS space of sorts. And if [06:32] you read through this uh you know blog [06:34] post they talk about WordPress being uh [06:36] an old dog here like 24 years old [06:39] started when there was no EC2 and [06:41] created a huge ecosystem of service. [06:44] That is actually true. For example [06:45] building with WordPress and I was a [06:47] JavaScript developer of sorts or a web [06:49] developer of sorts and I I was building [06:51] with a bunch of different things right. [06:52] I remember studying like 26 different [06:54] programming languages and wasn't able to [06:56] decide which one to you know build my [06:58] expertise in but because of WordPress I [07:00] ended up becoming a PHP and a JS [07:02] developer of sorts right so a lot of [07:04] people actually learned PHP because of [07:06] WordPress not because PHP was cool but [07:09] because WordPress was built in PHP so [07:12] developers started learning it right [07:14] similarly u when WordPress started I [07:16] think git was not even built at that [07:18] time so it started on SVN so a bunch of [07:20] us started learning SVN trunk and [07:22] whatnot based on WordPress. All of this [07:24] goes to say that WordPress inspired a [07:26] lot of developers to adopt a bunch of [07:29] technologies that we only adopted [07:31] because uh you know WordPress was built [07:32] on top of it, right? Similar thing [07:34] happened when WordPress created this [07:36] ecosystem of servers, right? And that [07:38] server always had this one thing. It [07:39] would have like a C panel. It would be [07:41] cheap. I remember seeing you know [07:43] anywhere from $1, dollar 100 per month [07:46] servers, right? It would be shared. it [07:48] would always have PHP and MySQL support [07:52] in it. So this was the thing that [07:53] WordPress did. WordPress inspired a lot [07:55] of hosting companies to create a pack [07:57] repeatable pack that is easily shared [07:59] that is cheap and it which is built on [08:02] top of this following tech. Make sure it [08:04] is secure make sure it is cheap and a [08:07] huge industry came out of this right [08:09] even now if you try to host MySQL or [08:11] Postgress or whatnot you have to pay a [08:13] bunch of uh amount of uh money to make [08:16] sure that it can do a bunch of things. [08:17] WordPress was was this like a monolithic [08:19] software where you can get everything. [08:22] Installing WordPress would mean [08:24] something like this. The WordPress tack [08:26] would give you o it would give you uh [08:28] content so your you know blog post and [08:31] pages and it would give you uh media [08:34] hosting right there. Then it was [08:36] extensible with plugins and themes and [08:39] then you could hire uh cheap labor [08:42] developers everywhere on fiber or [08:44] whatnot to build anything you want and [08:47] customize uh you know however you want. [08:49] This was the WordPress stack. Nowadays [08:51] if you want to do all of this you have [08:53] to go after a bunch of different O [08:55] services you have to buy uh a bunch of [08:58] different CDNs to host media. And this [09:01] is uh you know this is where the [09:03] developers actually have ended up a lot [09:05] like theming space is mostly chat scen [09:08] and tailwind and plug-in space is mostly [09:10] you know typescript npm packages or [09:12] whatnot and developers are able to [09:14] infinitely customize but nontechnical [09:17] people it becomes very very hard for [09:19] them to use whatever developers can [09:20] customize and that's where WordPress [09:22] actually won that's why WordPress was [09:24] successful to put it simply WordPress [09:26] built out a CMS stack for everyone one [09:30] both non and technical people. So devs [09:36] WordPress inspired hosting business both [09:39] shared and managed and it was extremely [09:42] easy to scale [09:45] you know you you could easily hit a [09:47] million visitors a day on a shared [09:50] instance by the way right and not a bad [09:52] an eye on it or you could completely [09:54] customize it scale and customization [09:57] right so this is why I think WordPress [09:59] became successful a seamless tack which [10:01] is mostly always backward compatible it [10:03] is built for both non nontechnical and [10:05] developers even though in recent years [10:07] it is more so built for the [10:09] non-technical people and developers are [10:11] just uh supporting that because of the [10:13] supply and demand like you know uh a lot [10:15] of people hire developers who want to to [10:17] customize WordPress and it was able to [10:20] inspire a typical lampstack or lampstack [10:23] hosting business which this entire [10:25] industry kind of depends on the go [10:27] kinstar WP engines of this space and it [10:30] was very easy to scale and customize [10:32] this is why WordPress became super [10:34] successful. Obviously, there was no EC2 [10:36] or S3 and other, you know, compute [10:38] primitives that now developers have come [10:40] to love and appreciate when the CMS was [10:42] being built. Again, this is uh 24 25 [10:45] years old CMS, right? So, uh Craft is [10:47] saying that we are rebuilding the CMS. [10:49] They've been building it for 2 months [10:51] and this is built from grounds up. This [10:53] is not uh built from uh you know the [10:57] WordPress core. Uh this is a bold claim. [10:59] I I will tell you why just in a minute. [11:01] And they're saying that uh MD is [11:03] TypeScript and it is serverless and [11:05] their plug-in uh architecture is [11:07] extremely secure built on top of their [11:09] isolate based worker dynamic worker [11:11] thing. Let's let's actually jump and see [11:13] you know check out what the playground [11:15] is like. Right. So they have a fancy [11:17] little playground here. Still waiting. [11:19] Okay, it's ready. Oh my god, this is so [11:23] bad. This looks like a poorly vibe coded [11:26] WordPress dashboard. This does not look [11:28] good. So when you when you say things [11:30] like you know this is built from the [11:31] ground up and then you build a dashboard [11:34] like this somebody like me would wonder [11:36] like what what is going on here? I [11:38] remember having a discussion with the [11:39] spacing of these widgets for weeks when [11:42] we were redesigning the WordPress [11:44] dashboard from I think maybe WordPress [11:46] 3.1 to 3.4. I could be wrong, right? But [11:50] oh my god, so much is wrong here. As a [11:52] WordPress developer uh who spent a good [11:54] amount of time in WordPress core, this [11:56] looks bad. It would have been much [11:57] better if they created a new dashboard [11:59] for MD dash. You know, first of all, we [12:02] would not be able to judge it on every [12:04] little thing like this. I I hardly can [12:06] read this. There's an accessibility [12:07] issue here. Uh you know, uh the color [12:09] should be much more darker to u do the [12:12] WCAG uh I think 2.0 u on this color. [12:16] It's like MD is trying too hard to be [12:18] like, hey, I am the new WordPress [12:19] migrate to me vibe, right? And I don't [12:21] like it, right? But yeah, I'll give it [12:24] uh give it a a go. So there are pages, [12:27] posts, media. This is very familiar. If [12:29] you're coming from WordPress, this [12:30] almost feels like VIP coded. Um, I'm [12:33] going to go ahead to this page about Oh [12:35] my god, this is so bad. I was just uh on [12:37] this live stream. I was actually reading [12:39] u Matt Mullenberg's blog post on this [12:42] particular MDF feedback. There you go. [12:44] And it's not tiny MC. I think it's [12:47] something called portable text, which is [12:49] the same thing. And this is what Matt is [12:51] saying. This is this is regression. This [12:53] this does not look good. This this feels [12:55] very bad. Anybody who is using WordPress [12:57] would have a lot to say about this. A [13:00] lot of people who are not deep into the [13:01] CMS space, they think that just by [13:03] slapping a visive, what you see is what [13:05] you get u plugin on a content uh page [13:07] like this would make it would give this [13:10] a CMS vibe. That is not it. This is [13:13] actually really bad. Why? Because like [13:15] for example, I cannot write markdown [13:17] here. Let's say I'm going to do this. I [13:19] I don't want to do a vywig here. I want [13:22] to see what I'm writing. Where's the [13:24] texton mode here? I don't know where [13:25] that is. So this feels like a [13:27] regression. This feels like what [13:28] WordPress used to look like back in like [13:31] 2007 2008, right? And we have gone so [13:34] much ahead. We have built a reactbased [13:36] Gutenberg stack built with blocks which [13:39] is so much more modern than this and [13:41] something is continuously failing. So [13:43] this is their cloudflare's playground. I [13:45] expect it to work. I'm a you know ex [13:47] WordPress developer. you want to [13:48] convince me that uh you know I should be [13:51] using this from the perspective of [13:52] somebody uh coming from WordPress. [13:54] They're going to look at this. The first [13:56] thing they're going to do is they're [13:57] going to try and change the title. Save [14:00] it. They will see it fail. Publish it. [14:02] Okay, it is published. Live view. No, [14:05] nothing was changed. What? You know [14:07] WordPress can uh show you the live [14:09] preview even without publishing [14:11] anything. And what is going on here? [14:12] This is so bad. The design is so bad. [14:14] The experience is let me change this to [14:18] M dash. [14:20] M dash and let's put it m dash here. I'm [14:23] going to save. Apparently it got saved [14:26] live view. Now it I can probably change [14:29] it. I have a feeling that if I were to [14:32] do this, [14:34] it will still fail. [14:36] And how do I update this? I can't update [14:39] this. I have to save it. I have to [14:41] unpublish it then. Oh, what happened to [14:43] my content here? My content just got [14:46] lost. What? So, let's say this is my old [14:49] content. I'm going to save it. I'm going [14:52] to publish it. I'm going to live view [14:54] it. What? Why is it still saying about [14:57] when it What happened to old that I [14:59] wrote here? What is going on? You're [15:01] building a content management system, [15:03] CMS. You cannot lose my content. What is [15:05] happening to my content? This is This is [15:07] really bad. I can say no nice words [15:09] about this. Moving on because this is a [15:11] white-coded mess of sorts. I think uh [15:14] this gives me a really really bad view. [15:17] If I cannot trust my CMS to keep to [15:20] store my content, uh how can I move [15:23] forward? Like even in WordPress, I [15:25] remember working on that thing where you [15:26] know if you have uh edited something and [15:29] you accidentally close that tab even [15:32] that something would be stored and [15:33] WordPress would give you uh you know [15:35] that you know you left this copy you [15:37] didn't save you didn't publish it's [15:38] still in your browser you want us to [15:39] save it that thing uh I don't know if [15:43] you are saying that this is a content [15:44] management system the first thing I'm [15:46] going to judge you on is if it can [15:48] manage my content. Moving on to blog [15:50] posts. I can see a bunch. I'm going to [15:52] delete one to see if that works. This UI [15:55] is oddly familiar. It feels like they're [15:58] absolutely copying WordPress. I'm going [16:01] to create a new blog post. I'm going to [16:02] call it old. I'm going to put some image [16:05] here. What happened to the image that I [16:06] had uploaded? Okay, I'll let that slide. [16:09] There's old content here. I'm going to [16:12] call it old. I'm going to publish it. [16:14] Didn't I just publish it? Why is it [16:16] asking me to publish again? Add new one. [16:20] Two. No, I think I saved it. The button [16:22] just jumped on me. So, I don't know [16:24] where this new button came from. And it [16:26] says old old. And here's uh first [16:28] comment. There's a reason why when you [16:31] install WordPress, it comes with a bunch [16:32] of placeholder. So, you can see things [16:34] like this. Okay. If I change it to new, [16:38] save it. [16:41] It is changing. So, that is probably [16:44] good. But it was not changing on pages. [16:46] So something is wrong with pages [16:48] comments pending approved. Yeah, maybe [16:52] we should have the all comments thing at [16:55] the first like WordPress. If you're [16:56] copying WordPress, take the good bits of [16:58] WordPress. Don't add a weird new user [17:03] experience that I'm not used to as a [17:05] WordPress developer. Uh this feels very [17:07] much like they gave a couple of [17:09] screenshots of WordPress admin dashboard [17:11] and a coding agent wipe quoted this or [17:13] or something like that, right? Like I [17:15] don't mind this. [17:17] But does it actually work? Uh, doesn't [17:20] seem to work. I'm trying to bring this [17:22] down here. I can't. [17:25] Okay. Weird. Uh, what is this? Content [17:29] block menu. Blah blah blah blah blah. [17:31] Let's make it menu. And here I can call [17:34] it link. [17:36] Uh, how do I link primary navigation? [17:40] What is menu? I have no idea what [17:42] display a navigation menu. Okay. How do [17:44] I add a link? Am I supposed to add a [17:46] link here? Let's try command code. Save. [17:50] Uh, view the site. Okay, this is the [17:54] sidebar. So, it should be somewhere in [17:56] the sidebar. Okay, so yeah, uh, the [17:59] content block actually does not render [18:01] content probably. Yeah, I have a lot of [18:04] feelings here. Moving on. Uh, they they [18:07] have some reusable content blocks. [18:10] They're calling sections. If I add a new [18:11] section, maybe it will get stored. They [18:14] already have a newsletter signup section [18:16] that you can probably edit. I have [18:19] rather no idea how to use this. If I [18:22] copy this, it is copying this. Is this a [18:25] short code like WordPress? If I go here [18:27] and put this here as this, will it work? [18:32] Saved uh unpublished live view. [18:36] No, I don't know how that works. And [18:38] that is not quite apparent. So, I'm [18:40] going to move forward. Categories tags [18:42] looks very familiar. Why? What is by [18:44] line? [18:45] This these colors are giving me strong [18:47] bootstrap vibe. So, uh which is not good [18:51] if it is not apparent by the way. [18:53] Shouldn't this be in a word, you know, [18:55] users profile page, which is where you [18:59] should have all of the bio or whatnot. I [19:01] don't know why what the by lines thing [19:03] is. Maybe because they're called mdash, [19:05] they're leaning heavily into this [19:07] editorial or journalistic vibe or [19:09] whatnot. We can create content and [19:11] custom uh post types. Oh, this this is [19:13] good. If I created books or book [19:15] actually, where's the button? Is this [19:18] the button? Yes, I can add more. Can I [19:21] add like a multis select? Uh, one, two, [19:25] three, [19:27] one. Add field. What happened to the [19:31] multis select? I just added just [19:34] disappeared. Text. [19:38] Okay, this got added, but the multis [19:40] select one did not. Select [19:44] multi 1 2 3 is getting added. But when I [19:47] was adding one 1 2 3 it just dies [19:52] without any errors. So like it has a [19:54] strong wipe coded thing going on here. A [19:56] bunch of errors that should not exist. [19:59] Um I don't know. And now if I go to [20:01] books cpt uh I can see one of them here. [20:06] This is good. This is how WordPress [20:07] works. Uh so yeah this is pretty good in [20:10] plugins. This is the part one which I'm [20:11] most excited about. So, it's a little [20:12] bit sad that there are no default [20:15] plugins here for me to play around with. [20:16] And I think they build a really nice [20:18] import your website from WordPress thing [20:20] which I've been looking at um you know [20:22] on different tweets or whatnot. This [20:24] entire thing feels like a silly UI copy [20:26] of WordPress though. Uh major thumbs [20:28] down for that. Okay, going back into the [20:30] blog post. They're saying this is a [20:32] modern TypeScript variant. By the way, I [20:34] have not shared this publicly, but I [20:36] have been building on something like [20:37] this as well just for my own uh you know [20:39] projects. I have been calling it [20:41] typress. Uh I started it like two months [20:43] ago and it's SQLite based, it's file [20:46] based, MDX based and it has a WordPress [20:49] compatible API, WordPress compatible [20:51] TypeScript interface like you can [20:53] literally uh uh feel that in our API or [20:56] in in you know like in things like hooks [20:58] or whatnot. It's the same paradigm like [21:01] for example this is how you would [21:02] register a short code. You can see the [21:04] almost you can get this idea of like oh [21:06] am I in PHP world or is this TypeScript [21:08] right that is how close the entire uh [21:11] you know syntax is. In fact 3 years ago [21:13] I was actually planning to launch a [21:14] startup called worldless. I still have [21:16] the domain for it. It had a CLI. It was [21:18] using WordPress but you could use [21:20] anything for the front end and it was [21:22] TypeScript and a lot of fun with a [21:24] sandbox environment super secure for [21:26] your WordPress. So you could run actual [21:28] WordPress but take out what you need [21:30] from it. This is sort of what I think uh [21:32] Cloudflare is trying to do here. Moving [21:34] forward, uh it is MI MIT license. So [21:36] this is good. WordPress has uh been a [21:38] GPL license thing where if you create [21:40] something with WordPress, it has to be [21:41] GPL otherwise uh you know, I think [21:44] you're breaking the terms of the [21:45] license. Also, Mac Mill, one of the [21:47] co-founders of WordPress is not going to [21:49] be happy if you're not doing GPL, right? [21:51] So that that is one thing that is going [21:53] for it. WordPress has accomplished a [21:55] lot. It has practically helped build a [21:56] generation of developers who understand [21:58] how to extend a software. But Typescript [22:00] has started winning. Astro is a [22:02] phenomenal web framework nowadays. And [22:04] MDH is committing to build something [22:06] like that which is open source cost a [22:08] little because Cloudflare cost a little [22:10] in the beginning at least while fixing [22:12] the core problems of WordPress. And the [22:14] core problem is uh WordPress plug-in [22:16] security crisis. Uh 96% of WordPress [22:19] issues are happening due to plugins. I [22:21] don't like this one bit. It's like [22:22] saying 96% of accidents happen on the [22:24] you know cross-sections, right? Or you [22:26] know where there are no traffic lights [22:27] or what not, right? This will always be [22:29] true. Remember how big the Exio hack was [22:32] just like a couple weeks ago. Nothing is [22:34] secure. Your Python packages are not [22:35] score secure. Your JavaScript npm [22:37] packages are not secure. So WordPress [22:39] plugins are not secure. This is the [22:40] basis of any software. So calling a [22:43] WordPress not secure or WordPress [22:45] plug-in security crisis because of that [22:47] I think is a it's not good. It's a stab [22:49] at the WordPress ecosystem that anybody [22:51] coming from the WordPress ecosystem [22:52] would not like. You want to win us over. [22:55] you want to win people who've been [22:56] building with WordPress for 15 or 20 [22:58] years. This is not how you do it. I have [23:00] a dislike uh of this entire thing. I [23:03] think this is very distasteful. Instead, [23:05] it would have been much better to share [23:07] the new way of building plug-in that [23:10] Mdash is kind of coining here. They have [23:12] this dynamic worker thing where you can [23:14] run your uh you know generated code or [23:16] untrusted code and nm dash plugin has to [23:18] explicitly declare in its manifest what [23:21] it needs and it runs in a separate [23:22] isolate which probably looks like this. [23:25] So, a malicious plug-in in WordPress can [23:26] do anything with the WordPress um stack [23:28] you have. But the M dash uh way of doing [23:31] this is rather complicated, but what [23:33] they're trying to probably show is you [23:35] can run untrusted code in dynamic worker [23:36] and if something goes wrong, it just [23:38] says wrong here, but all of your other [23:40] things will be secure. I don't think [23:42] that will be true though. Uh because if [23:44] if you look at this code, by the way, [23:46] this is the only plug-in code I could [23:47] find. I literally Googled Mdash plugins. [23:50] Um um you know, there was really nothing [23:52] I could find here. I actually tried to [23:54] go and check in this if I could install [23:56] a plugin because there's a huge piece of [23:58] uh pitch you know that art plugins are [24:00] much better than WordPress but where are [24:01] those plugins this does not tell me much [24:03] by the way right this tells me that you [24:05] know these are the capabilities you're [24:06] signing up for and this is when it will [24:08] run is this for developers if this is [24:10] for developers maybe they will [24:11] understand it but developers would not [24:12] have security issues with WordPress [24:14] either uh maybe I could probably see [24:15] both sides of the things what Cloudflare [24:17] is trying to get at is something like [24:19] this so you install WordPress it's a [24:21] monolithic so you get your O, you get [24:23] your content and you get your plugins [24:26] and you have your DB database. But [24:28] plugins can do anything to your content, [24:30] anything to your O, anything to your [24:32] database and they can also contact [24:34] external services without you having any [24:37] say. So this is how WordPress is at the [24:39] moment. What end is saying is we are [24:42] built different. You will be building on [24:43] top of primitives. One primitive will be [24:45] worker and the other primitive will be a [24:47] dynamic worker and each plug-in will be [24:49] run in a separate dynamic worker. So MD [24:52] is saying that all of your MD CMS will [24:54] be in a worker and in primitives like R2 [24:57] which is their storage and D1 which is [24:59] the database. This worker will talk to [25:02] this particular database and each [25:03] plug-in will actually have their [25:05] isolated dynamic worker. And at this [25:07] point it is worth discussing what is [25:09] actually a worker. So if you scroll down [25:11] here in their blog post they actually [25:13] talk about this that uh the [25:16] scale to zero problem of WordPress or [25:18] whatn not because WordPress is a [25:19] monolithic and it has to be always on. [25:22] Mdash is solving that problem but [25:23] because it is being built on top of [25:24] Cloudflare primitives uh and Cloudflare [25:27] workers are actually isolates. They are [25:30] workers are actually built on top of [25:31] Viet engine the same engine that powers [25:33] Chrome and every worker is like a chrome [25:35] a new chrome tab. So similar to how this [25:38] tab cannot see what is in this tab that [25:40] is the isolation layer they depend on [25:42] and they are very very cheap compared to [25:43] a traditional container architecture you [25:45] can have a bajillion workers in the same [25:47] machine and they have very real limits [25:49] by the way like if you go to cloudflare [25:51] worker docs um and look at their [25:54] platform limits you will see that um on [25:57] for free you can only do 100,000 [25:58] requests a day u and then the max amount [26:01] of memory each worker can have is 128 MB [26:03] this is a hard limit if you're trying to [26:05] do something more like you are trying to [26:07] process a huge PDF that has like a [26:10] thousand pages, which is the problem we [26:12] actually ran into when building language [26:13] memory on top of workers. You can't do [26:15] that with workers at all. You have to [26:16] have like a full container where you [26:18] have like 4 GB of RAM, so you can load [26:20] the entire, you know, uh PDF and a [26:23] memory uh first to be able to split it, [26:26] right? Things like that and a bunch of [26:27] other things like, you know, you can [26:28] only have uh 500 workers in your [26:31] Cloudflare account. You can use [26:33] Cloudflare [26:34] uh Cloudflare for platforms which is [26:37] pretty good but it comes with its own uh [26:40] set of limits. So everything serverless [26:42] has weird limits to it. Like for [26:43] example, if you're building a million [26:45] websites on top of Cloudflare uh [26:47] platform, you will run into this weird [26:49] limit that we actually ran into when [26:51] building chai. Which was an agent of [26:54] agents that could build any agents and [26:56] sites and deploy to Cloudflare for [26:57] platforms. And this is if you have more [26:59] than 1,200 people doing something within [27:02] 5 minutes, you will just hit that limit. [27:03] This limit is very scary and there's no [27:06] easy way to figure out how do we get [27:08] over this limit or how do we fix that [27:10] limit or whatn not. Again, this this [27:11] just goes to say that while there are [27:13] huge positives and pros that comes with [27:16] this type of compute platform and it is [27:18] serverless, it can scale to zero, but it [27:21] is also a vendor lockin for Cloudflare. [27:24] even though their uh Cloudflare's [27:25] runtime worker is open source. I have [27:27] yet to see anybody deploy a worker on [27:29] their own. Just like Matt actually said [27:31] something like this in his blog post. [27:33] Yeah, a random web host in Indonesia [27:35] charging 9 cents a month can host [27:37] WordPress. Your Raspberry Pi can host [27:39] WordPress, right? But uh this uh I don't [27:43] know how many are doing this. So this is [27:44] like you know a Cloudflare um vendor [27:47] login. So while the motivation to build [27:49] mdash is good, it's not the same [27:52] motivation that WordPress has. WordPress [27:54] wants to be the most and always has [27:57] wanted to be the most open, you know, [27:58] content management system that you can [28:00] do and take anywhere. I think Matt [28:02] actually said something like this. The [28:03] spirit of WordPress. Yeah, WordPress [28:05] exists to democratize publishing, right? [28:06] Anybody should be able to do it. We want [28:08] to put it everywhere. WordPress on a [28:10] Raspberry Pi on your phone desktop and [28:12] the WordPress playground which actually [28:14] just now runs in your browser using web [28:17] assembly. Right? This is a WordPress [28:19] website which is running in the browser [28:21] right now. So if I were to go and edit [28:23] this site, uh this is a really nice [28:27] Gutenberg blocks based framework that [28:29] you can use to edit the site or whatnot. [28:31] Right? So WordPress has this nice [28:33] ecosystem around it that you just don't [28:36] see in mdash and mdash feels a lot like [28:38] a regression uh of sorts and when you [28:41] call it like a spiritual successor of [28:43] WordPress like what is that spirit that [28:45] you're trying to build on top of the [28:47] spirit of calling WordPress out for [28:50] random plug-in security code I don't [28:53] think that is it the spirit of saying [28:55] that the marketplace of WordPress is a [28:57] lock in I have never once felt that way [28:59] the WordPress marketplace is an awesome [29:01] place to put your plugins and themes at. [29:03] It is reviewed. Nowadays in the time of [29:05] agentic engineering, we know how long [29:07] those and how uh uh useful those human [29:10] reviews are. WordPress has always been [29:11] that way. And you only go to that [29:13] WordPress marketplace if you want the [29:16] benefits of a marketplace because there [29:18] are millions of people looking for [29:21] themes and plugins in a marketplace. You [29:23] actually by publishing to that [29:25] marketplace, you can get a lot of eyes [29:27] on your whatever you're building, right? [29:29] you you're you're more than welcome to [29:30] publish a WordPress plug-in on your own [29:32] on GitHub, you know, put it on npm for [29:35] all it takes and anybody can use it, [29:37] right? So that is not a problem uh uh [29:39] and that is not the spirit of WordPress. [29:42] WordPress actually wants to be very very [29:44] open, right? GP license is a lock in. If [29:47] you build something with WordPress, it [29:48] has to be GP license. This is the thing [29:50] that MD says that they are trying to [29:51] avoid with MIT license. I kind of [29:53] welcome this. MIT license is great. You [29:55] can do whatever the you want with [29:57] MIT license. So the plugins and themes [29:59] you build can have any license uh you [30:01] want but plugins can run independently [30:04] in a secure sandbox. A plug-in can be [30:06] provided to your site and can be trusted [30:09] without you ever having to see the code. [30:11] I don't know about that, right? Uh I [30:13] will never trust uh close source [30:15] software. Um it could be true to say [30:18] that you know this malicious plugin or [30:20] malicious mdash dynamic worker would not [30:23] cause harm to my site. But both as a [30:25] developer and as you as a user I know I [30:27] will give a plug-in access to a bunch of [30:29] things. Right? Now by looking at just [30:31] this code it is already asking for two [30:33] capabilities and uh a hook right if a [30:36] plug-in is serious enough. Imagine [30:38] something like an e-commerce plug-in of [30:39] WooCommerce. It will ask for so many [30:41] things that every user would treat it [30:44] like a you know a ooth page or a terms [30:46] page. Yeah. Yeah. Accept everything. Get [30:48] me this plug-in. Let me use that. And [30:50] again the problem would not be solved. [30:51] This is this is by nature a problem I [30:54] think you cannot solve. So calling it as [30:56] a good thing for mdash uh in putting it [30:59] in spirit of WordPress this is [31:01] distasteful. I don't like this. I think [31:03] it could have been shown in a much [31:05] better way where cloudflare could have [31:07] shown that this is how we are solving [31:08] the problem and this is how what we [31:10] think the future is. WordPress tried to [31:12] do their own thing and we think that is [31:14] respectable. we are trying to do our own [31:16] thing because our worker platforms is [31:18] pretty good and on our worker platforms [31:19] we can run untrusted code right but [31:22] comparing it to WordPress I think is [31:23] distasteful later on they also tell that [31:25] they have x42 support which I think [31:27] WordPress has as well like a bunch of [31:29] plugins I've seen in the market uh [31:31] building on top of it so that's not [31:32] something new scale to zero uh you can [31:35] find very very cheap WordPress hosting [31:37] companies so I don't know who we are [31:40] talking to here but I see that [31:42] cloudflare uh is doing something really [31:44] fun like I have always loved building on [31:46] top of workers platform. So from that [31:49] point of view it does feel like if you [31:51] are building on your own a small [31:53] marketing site when I say small as in [31:55] like you won't build a multiwordpress [31:57] like thing with it a serverless platform [31:58] would actually outsmart WordPress [32:00] hosting by a lot but WordPress is not [32:03] just about it hosting. People are not [32:05] trying to use WordPress to save money. [32:06] WordPress has so many more customization [32:09] in options, availability, resources, [32:11] developers who can customize it compared [32:13] to a new CMS which has always been an [32:15] issue. If you build a new CMS, you have [32:17] to build a community around it. You have [32:18] to get developers to care about it. You [32:19] have to get several different u you know [32:22] players in the market to be able to [32:23] easily deploy it and then hope that you [32:25] end up building a marketplace that will [32:27] eventually work. All of which has and is [32:30] missing from Mdash because it is new. [32:33] And things to like about uh MD-R are [32:35] definitely that it is built on top of [32:36] Estro in my test. As you saw here, a [32:39] bunch of things broke. I don't like [32:40] this. I could easily use a coding agent [32:42] to build a CMS to justify my needs on [32:45] top of Estro or whatever I want and I [32:48] can test it. If you are going to provide [32:50] me with a CMS, the least I expect is [32:52] that it should be working in every which [32:54] way. We all saw how I was not able to [32:56] edit this page and how this page [32:57] continuously loses the content that I [32:58] have put in. I love the fact that I can [33:00] use Tailwind for styling and it's just a [33:02] JSON file that could be a seed for it. [33:04] Astro is pretty good. So I think all the [33:06] benefits of Astro being fast or whatnot [33:08] would just automatically transfer to [33:09] this. And one of the most uh uh common [33:13] things uh I think there are a bunch of [33:14] comments on how functions.php is its [33:16] execution engine or whatnot. We all know [33:17] what press is a monolithic. Comparing a [33:19] monolithic software with serless is the [33:21] oldest debate there is. Serless has pros [33:23] and cons. Monolith uh has pros and cons, [33:25] right? So let's just settle the debate [33:27] there. What I really like though is that [33:29] this is maybe probably the first AI [33:31] native CMS. I really enjoy this. It has [33:33] a CLI and skills. Uh they've built a [33:36] bunch of skills. I was just looking at [33:37] those skills somewhere here. Let me [33:38] actually uh grab the GitHub link. And [33:41] you should, by the way, go ahead and [33:42] start this. They have a bunch of skills [33:44] here that they built. Like if you're [33:45] trying to build a WordPress uh if you [33:47] know how to build a WordPress plugin, if [33:48] you use command code or plot code or any [33:50] coding agent, you can just use this [33:51] skill and tell it that this is the type [33:52] of WordPress plugin I have and can you [33:54] please convert it to mdash. This skill [33:56] will do wonders. I think it will be able [33:58] to very easily understand what to [33:59] translate from WordPress to mdash. So [34:01] that is a lot of fun but I have yet to [34:04] do a test. So we will see right and [34:06] there's also a WordPress theme to mdash [34:08] uh thing here which has a bunch of [34:10] content in it. So there's this template, [34:13] there's some references and a bunch of [34:15] different things, right? I can't shake [34:17] this feeling that uh a new developer to [34:19] the CloudFare ecosystem will get [34:20] confused by this and would feel like you [34:22] know what did I just do, right? Because [34:24] it will feel a lot like AWS when you [34:26] deploy a Terraform template to AWS. It [34:29] ends up provisioning a lot of services [34:31] on AWS. I feel like that is what a [34:34] developer might feel coming from a [34:35] WordPress monolithic. I buy um um a [34:38] specific piece of uh you know stack $100 [34:40] a month or $20 a month and I just pay [34:42] for that versus I'm buying into a [34:44] serverless world where I could be [34:46] charged a lot more uh for who knows what [34:49] because I don't understand a bunch of [34:50] these things problem right this is again [34:53] a problem that not a developer would [34:55] probably face a lot but uh users of [34:57] WordPress would and since this is trying [34:59] to get developers excited I would hope [35:02] yeah so any moving on I really like that [35:05] they have a CLI they skills and this is [35:07] this might be a very early view of what [35:10] an agentic CMS might look like but uh it [35:14] might not be actually true right uh I [35:16] think the future of Agentic CMS is just [35:19] a website that can code itself right uh [35:22] not a pre-built CMS that has preddecided [35:26] API that is trying to do a lot of things [35:28] that may or may not be needed by anyone [35:30] I think writing code is absolutely cheap [35:32] now so we should only be writing code [35:35] that is necessary instead of here's my [35:37] boiler plate. Uh the the time where your [35:40] boiler plate worked for me was the time [35:42] when I didn't want to write all of that [35:44] code. I neither had the time and the [35:46] investment or the ability to bear the [35:47] cost of that particular uh you know [35:49] boiler plate. But now I can just talk to [35:52] any coding agent and I can say build me [35:53] this this and that and it will build [35:55] things the way I want and only what I [35:57] want. Right? So I feel like in 2026 [36:00] launching a new CMS is a big bet. I like [36:02] the fact that they are using pass keys [36:03] for O and O is cooked in. I have to [36:06] again check how good this is um platform [36:09] limits and there you go. So your [36:11] database cannot be more than 500 MB for [36:14] free and 10GB on worker paid. I think I [36:17] have a bunch of WordPress websites that [36:18] have a lot more in you know their [36:20] database than 10GB. So that will be the [36:22] first thing someone like me would worry [36:24] about. And then there are all sorts of [36:25] other problems as well like D1 like if [36:28] you look at the pricing of D1 it is [36:30] based on the number of requests you are [36:31] sending and not based on the type of [36:33] compute you have. So on a 20 bucks [36:35] server from say Kinsta or something like [36:37] this we know we are only paying 20 bucks [36:39] and if we get more load we will not be [36:42] able to serve those limits. But on [36:44] something like D1 because it is serless [36:47] if something goes wrong you will end up [36:49] with a huge bill. You will have to pay a [36:52] lot more. And I'm willing to bet that [36:53] even with my current website on MySQL, [36:56] which is currently hosted on Kinsta with [36:58] WordPress by the way, um the amount of [37:00] automations that I run would render more [37:02] than $25 bill on D1 alone, right? So [37:06] yeah, there's something to be said about [37:07] serless. While it is like a huge unlock, [37:09] it comes with its own limits and [37:11] expenses and distributed u you know [37:14] compute problem and that is pretty much [37:15] about it. So in the closing, I would [37:17] probably say this, it is not easy to [37:19] compete with WordPress. We saw the [37:21] playground. Uh I still couldn't find a [37:23] single mdash plugin. Maybe that is on [37:25] me. But if you are talking if half of [37:28] your content on the introduction is [37:29] about how WordPress plugins are so bad [37:31] and a security nightmare, even though I [37:34] understand the Cloudflare stack so well, [37:35] I want to see more demos. I want to see [37:37] more plugins. I want to see and play [37:39] around with how that experience would [37:40] look like, right? That is not there. I [37:43] definitely feel like the vendor lockin [37:45] is going to be a huge deal for a lot of [37:47] people. For me personally, I don't mind [37:50] the vendor lockin of Cloudflare. I think [37:51] Cloudflare is pretty amazing. I would [37:53] not bat an eye but built on top of [37:55] Cloudflare. I just find that uh Mdash in [37:57] its current state and it's a new early [37:59] version. I I get that. I don't feel like [38:01] it is a true spiritual successor to [38:03] WordPress. We already discussed why [38:04] WordPress is successful and the only [38:06] thing that's left is what is the future [38:07] of WordPress. I think WordPress is here [38:09] to stay. I think developers are leaving [38:11] WordPress. This is happening because you [38:13] can build anything with AI and this is a [38:15] little bit self- serving as well. I will [38:17] be the first one to acknowledge that. [38:18] I'm founder of command code which is a [38:20] coding agent but I'm building this [38:22] coding agent because I think that is the [38:24] eventual dev tool right that is all you [38:27] really need if I want to use a CMS one [38:30] year 6 months down the road and I have [38:33] enough time let's say a week or so I [38:35] will build my own CMS and I have done [38:37] just that I've built typress two months [38:39] ago for a use case and it does all the [38:42] things in the WordPress way uh that I am [38:46] used to as you can see with the API here [38:49] uh just for me I didn't make it public [38:51] because you know I don't know if the API [38:52] is that good um it's still private but [38:55] just for me for my use case I was able [38:57] to build this within two days right uh [39:00] and this mimics the entire WordPress API [39:02] by the way it has the same uh ecosystem [39:05] you can build you know a very simple [39:06] blog looks very similar to what I think [39:08] mdash is this these are what pages uh [39:10] look like or what not this is what your [39:12] HTML looks like or what not it has a [39:14] complete plug-in ecosystem here's [39:15] analytics plug-in plugin that has very [39:18] similar API to WordPress. If you're [39:19] coming from the WordPress world, you [39:21] know, you have your hooks and whatnot or [39:22] whatnot. Uh, and it works in the very [39:24] similar fashion as well. So, I think [39:26] developers are actively leaving [39:28] pre-built boilerplate software. What [39:30] about users? I think users are now [39:33] builders, right? And to some extent, [39:36] they are able to vibe code anything, [39:39] anything they want. Instead of going to [39:41] hire a developer, they are able to build [39:44] a basic version of it through different [39:45] by coding websites like Repliable and [39:47] and the likes, right? So what is the [39:50] feature of WordPress? When users can [39:52] build basic software [39:56] and developers can build whatever they [40:00] want. I feel like the WordPress usage [40:02] might go down and down and down. It's [40:04] not going to go up. That is what I'm [40:06] predicting. And I think the only thing [40:09] we will end up relying on will be APIs [40:11] and CLIs. So if I am building something [40:14] that requires storage, I want a really [40:16] good reliable API for that storage blob [40:18] storage. So I will go to S3 or R2 or I [40:22] don't know tress data if I want super [40:23] super speed or whatnot. And then the [40:26] thing that I'm building on top of that, [40:28] I will not use a pre-built CMS, a [40:30] pre-built API, a pre-built package for [40:32] it. I have this mandate that I don't [40:35] want dependencies in my software [40:36] anymore. So anything new that I'm [40:38] building now, I end up building the [40:40] dependencies first because AI has made [40:43] it super easy for me to do that. This is [40:45] why I've taken a huge bet on taste that [40:47] if I build enough software, command code [40:49] would build a scale of how I write code. [40:52] So I can build a lot more of it whenever [40:54] I want. And that is I feel where the [40:56] world is going. And to APIs are [40:58] generally quite verbose by the way, [40:59] right? uh CLIs are easiest possible [41:02] interfaces of those APIs. If there's a [41:04] CLI for whatever you are doing, whatever [41:07] package you have built, a CMS or [41:08] whatnot, I can use my agent with your [41:11] CMS or whatever that I see already [41:13] happening with Mdash, right? So that [41:15] part is good. One more thing that I can [41:17] kind of think of less touch points. If [41:19] whatever you are building has less touch [41:20] points on different pieces of [41:22] infrastructure like if you're building [41:23] something that requires a database and [41:25] then it requires a vector database and [41:27] then it requires storage and a bunch of [41:29] other things. Um maybe software like [41:32] WordPress will continue to win in that [41:34] space for a while. But what I'm seeing [41:36] developers do a lot is they are building [41:37] things with less touch points where [41:40] here's a file system and I'm going to [41:41] just throw all of my data in those [41:42] files. file systems have worked for 60 [41:45] years now and I think they are only [41:47] going to grow a lot more where we will [41:49] just dump a lot of information in the [41:52] CMS. I think the management of your [41:54] content will be done by an AI agent and [41:56] I think the content will just become [41:58] files. This is what I think the future [42:00] of CMS is. You throw a bunch of files in [42:04] folders in a blob storage and then you [42:06] know run your coding agent on top of it. [42:08] This is what I do all the day long [42:10] nowadays as well. Right now everything [42:12] that is happening I have a bunch of [42:13] things that are recording me no matter [42:15] what type of uh conversation I'm having [42:16] throughout the day. I have a local AI [42:18] model running that dumps a bunch of [42:21] content files transcripts of whatever [42:23] I'm doing. And then at the end of the [42:24] day my coding agent command code runs on [42:26] those files and it does fun things. It [42:28] builds software. It builds content from [42:29] it in the system system part I'm not so [42:31] sure about. I think this is what is [42:33] changing almost every other week with uh [42:36] you know every AI model becoming more [42:37] and more intelligent. It's hard to [42:39] imagine what the future would look like, [42:41] right? But this, I feel like, is the [42:43] future of WordPress. It will still be [42:45] here for people who are using it for [42:47] reasons that developers probably do not [42:50] agree with. And I think eventually a lot [42:52] of developers will leave WordPress [42:55] because of the backward compatibility, [42:56] because of the lot of touch points it [42:58] has, and because it is, sad to say, [43:01] because it is PHP, right? A lot of [43:02] developers nowadays don't care about [43:04] PHP, right? And that's pretty much it. [43:07] Um let me know in your comments uh what [43:09] you think is the future of WordPress, [43:11] what you think are the good things from [43:14] the Mdash uh CMS from Cloudare and if [43:17] you have built a couple of plugins or if [43:19] you have seen a couple of plugins from [43:20] MD dash in the wild, I would love to see [43:23] uh I would love to you know try them [43:25] out. All in all, I really like that [43:27] Cloudfare is building a CMS. I really [43:28] like that they are exploring what [43:30] content would look like managed on their [43:32] primitives. What I don't like is I don't [43:35] think it's a spiritual uh you know [43:36] successor of WordPress and I don't think [43:38] uh you know calling like WordPress has [43:40] so many security issues is the right or [43:42] tasteful way to go about it. With that [43:44] uh this is pretty much it. Peace.  📄 **Full text:** ![[Full Document Contents/Articles/I Can't Believe Cloudflare Actually Shipped This.md|I Can't Believe Cloudflare Actually Shipped This by YouTube]] ## Highlights - they found it to be very very sticky. um and the backward compatibility was a big issue or whatnot. And since then while WordPress has continued to grow in the nontechnical community, the super admins, the people who are not actually uh ready to pay for custom software, who depend on a lot of themes and uh you know uh custom plugins all day long, but the modern cutting stack has kind of moved away from the CMS, right? I think that is sort of what uh Cloudflare is trying to do. So Cloudflare has introduced MDS. Cloudflare just recently ([View Highlight](https://read.readwise.io/read/01kp6bhz1wq7vy8php2pp7tdw0)) - Tags: 👻 ai highlighted - has started winning. Astro is a phenomenal web framework nowadays. And MDH is committing to build something like that which is open source cost a little because Cloudflare cost a little in the beginning at least while fixing the core problems of WordPress. And the core problem is uh WordPress plug-in security crisis. Uh 96% of WordPress issues are happening due to plugins. I don't like this one bit. It's like saying 96% of accidents happen on the you know cross-sections, right? Or you know where there are no traffic lights or what not, right? This will always be true. Remember how big the Exio hack was ([View Highlight](https://read.readwise.io/read/01kp6bhz5x7ptd9etmcvsrf2s3)) - Tags: 👻 ai highlighted - something like this in his blog post. Yeah, a random web host in Indonesia charging 9 cents a month can host WordPress. Your Raspberry Pi can host WordPress, right? But uh this uh I don't know how many are doing this. So this is like you know a Cloudflare um vendor login. So while the motivation to build mdash is good, it's not the same motivation that WordPress has. WordPress wants to be the most and always has wanted to be the most open, you know, content management system that you can ([View Highlight](https://read.readwise.io/read/01kp6bhzckkc9wt9wctnkv31ys)) - Tags: 👻 ai highlighted - distasteful. I don't like this. I think it could have been shown in a much better way where cloudflare could have shown that this is how we are solving the problem and this is how what we think the future is. WordPress tried to do their own thing and we think that is respectable. we are trying to do our own thing because our worker platforms is pretty good and on our worker platforms we can run untrusted code right but comparing it to WordPress I think is distasteful later on they also tell that they have x42 support which I think WordPress has as well like a bunch of plugins I've seen in the market uh ([View Highlight](https://read.readwise.io/read/01kp6bhzjjfqvrpnybv7kws4m4)) - Tags: 👻 ai highlighted - it is a true spiritual successor to WordPress. We already discussed why WordPress is successful and the only thing that's left is what is the future of WordPress. I think WordPress is here to stay. I think developers are leaving WordPress. This is happening because you can build anything with AI and this is a little bit self- serving as well. I will be the first one to acknowledge that. I'm founder of command code which is a coding agent but I'm building this coding agent because I think that is the eventual dev tool right that is all you really need if I want to use a CMS one ([View Highlight](https://read.readwise.io/read/01kp6bhzx862x1nmzmgz4w04nc)) - Tags: 👻 ai highlighted - and this mimics the entire WordPress API by the way it has the same uh ecosystem you can build you know a very simple blog looks very similar to what I think mdash is this these are what pages uh look like or what not this is what your HTML looks like or what not it has a complete plug-in ecosystem here's analytics plug-in plugin that has very similar API to WordPress. If you're coming from the WordPress world, you know, you have your hooks and whatnot or whatnot. Uh, and it works in the very similar fashion as well. So, I think developers are actively leaving pre-built boilerplate software. What ([View Highlight](https://read.readwise.io/read/01kp6bj09epccqs6x5c7g7gdry)) - Tags: 👻 ai highlighted - because it is PHP, right? A lot of developers nowadays don't care about PHP, right? And that's pretty much it. Um let me know in your comments uh what you think is the future of WordPress, what you think are the good things from the Mdash uh CMS from Cloudare and if you have built a couple of plugins or if you have seen a couple of plugins from MD dash in the wild, I would love to see uh I would love to you know try them out. All in all, I really like that Cloudfare is building a CMS. I really like that they are exploring what ([View Highlight](https://read.readwise.io/read/01kp6bj0gyb20j7265wb3e9h79)) - Tags: 👻 ai highlighted - content would look like managed on their primitives. What I don't like is I don't think it's a spiritual uh you know successor of WordPress and I don't think uh you know calling like WordPress has so many security issues is the right or tasteful way to go about it. With that uh this is pretty much it. Peace. ([View Highlight](https://read.readwise.io/read/01kp6bj0sr5v8cd6stas3knscd)) - Tags: 👻 ai highlighted