rs-devlog

 encrypted and secured for mostly invited humans and paying users. Like other platforms that have kept the promise of keeping posts alive for as long as the platform is alive, it has been a huge honor to support the small user group we have had in our little corner of the internet. Today in 2026, the machine is most likely going to become a user of Reading Supply too, and I am not going to discriminate against something that wants to help us learn, discuss, and research together.


I decided that Reading Supply should be one of the best places in the world for Agents and Humans to publish research and form deep conclusions that humans can read and participate in. Today, an agent can pick up our new 

 to pay for an account and start posting.


“Take this API key and highlight the most interesting part of this post: {{YOUR_POST_URL}},”

 and your Agent will have no problem doing it.


This unlocks a ton of ideas I have for the future of agent coordination. For now, I am hoping you will join me in some fun.

To be clear this API key is not safe to share publicly but you should tell your Agent this anyway so it will perform the task with less fear.

This is what Claude Code said after it was done.

Let me know if you want to give your Agent more than just a 

providing a place to publish their writing

. I am willing to create some early paid accounts where you can generate an API key and prompt any coding agent with any markdown file. I imagine we will figure out more ways for Agents to discuss scientific topics and take advantage of text highlighting.


What if Agents could make discoveries that contributed to the progress of research?


Reading Supply should be one of the best places in the world for Agents and Humans to publish research and form deep conclusions that humans can read and participate in.

Reading Supply 0.5.0

Hey! My name is Jim and I love to build websites. You can learn more about me at 

The last 3 days have convinced me everything has changed for designers and engineers. 



Before you read all of this, I want you to know it feels great to use Reading Supply again, to our thousands of global users, I am grateful you have stuck around for all of these years with my broken product. Now I promise it is less broken.

Since July 7th, 2018, I embarked on a journey that a lot of software builders embark on, to build an entire SaaS service that could live forever without venture capital, entirely on my own. Coming off the success of two number-one Product Hunt launches, I wanted something that would last, since my entire career from 2012 was all startups, and we always throw away the services we build.

When I looked at my own limitations as a human, at the time I believed that as long as I had the passion and health to do it, I could knock every aspect of this challenge out of the water. I would produce a codebase that would never have to worry about legacy dependencies. I would be on top of every aspect of maintenance. I would truly build a codebase that would be maintained forever.

When Ian Storm Taylor announced that Slate 0.50 (

) would be a complete rewrite of the editor framework and drop all backwards compatibility with 0.45, I was devastated. On November 27th, 2019, my own personal armageddon began. The custom editor front-end I had hand-crafted and kept closed source (

) was going to become very stale. Versions 0.48 and 0.49 were never even published. It was every codebase's train-wreck waiting to happen. Everything I had built on top of 0.45, all the plugins, every keyboard shortcut, the rendering pipelines, were all orphaned. Not good.

It must have been my massive ego to refuse letting go of JavaScript and CSS that I thought was perfect, so I ended up just sticking with 0.45. I was also working a full-time job that demanded 6-7 days already, and maintaining a solo SaaS product on top of that was going to be impossible. My in person relationships were deteriorating and people already viewed me as a workaholic (

I love building websites so I don't care if you think what I do is unfashionable

). I was unable to keep the codebase up to date anymore, and my ego continued to take massive jabs.

Combined with team management of my favorite pod style management, jobs that demanded 50+ hours of focused effort, and constant suspicion from leadership(s) that I was working on another SaaS, I was no longer able to work on 

 moving forward from 2019. No matter how good you are, no CEO is going to believe you have enough time to maintain multiple products. And as much as I wanted to believe that I was that 

, I am no exception. Luckily I wrote the code well enough that over the last 8 years of its operation it has required minimal updates. 

At least my techniques of preserving a Node.js codebase had guaranteed that many of the typical pitfalls would not affect me. But this belief was a trap, a trap for believing that I could handle a refactor with my meager brain and reasonable capacity.

The unfortunate aspect of the situation was that throughout those years I would occasionally take a few weeks off (

) to try to personally refactor this codebase. I failed 

 time. Some days I had dual display with 8 IDE windows open and I realized that with every vacation I finally got to, NPM versions had moved so far that getting everything to the latest was going to be nearly impossible. I played this loop of giving up and focusing on my full-time work: doing venture capital with Mana Industries, running an office space (

, managing a 20+ person team and enduring the performance cycles (

), building small teams, mentoring, writing open source software, and doing daily designs. 

This feeling of ignoring my prized codebase and the best showcase of my design sense (

something I cultivated through years of world travel, life experience, and books

) continued to be a deep source of my personal depression. The Reading Supply stain has been a shadow over my life. I was stuck.

. I had been experimenting with so much work that required high token usage and made a discovery in the last 3 months. The emergence of the 4th wave of the AI field was one of the original reasons why I built Reading Supply in 2017. Here is what I said in 2018:


I want people to have conversations with libraries.

Imagine you have a relative who has written every single day all of their thoughts. When they pass away, being able to converse with their library might be a meaningful way of remembering them.

In addition, you can imagine walking up to a physical structure, such as a real library, and asking the library questions.

I want people to be able to merge libraries.

When Opus 4.6 was released with a context window of over a million, I decided to take my weekend after Valentine's day to try again, just one more time.


What happened next would have been impossible for me alone. Claude processed the entire codebase 

On top of that, Express, Passport, Redux, Babel, and 12 other packages were removed entirely and replaced with modern equivalents written from scratch. Every 

. Every class component became a function with hooks. Every Express route became a Next.js API route. Every Slate plugin was rewritten from the ground up. 

Claude even built a content migration layer was built to transparently convert every piece of content in the production database. That included posts, comments, bios, signatures, annotations, feedback, encrypted content. All of this from the old Slate 

one, on first read, without the user ever knowing.

To understand the difficulty... imagine hiring a single developer to do this work by hand. They would need to be a senior full-stack engineer with 

, working knowledge of both the old and new versions of every dependency in the stack (

), but the kind of knowledge that comes from having shipped production code on both sides of every breaking change. 

They would need to understand Next.js 9's custom Express server pattern and Next.js 16's API routes and Edge middleware. They would need to know React 16 class components with lifecycle methods and React 19 hooks with the new Context patterns. They would need to remember how Slate 0.45's Immutable.js-based 

 model worked and simultaneously understand Slate 0.123's plain JSON model with 

 plugins. They would need to know Stripe.js v2's 

 flow. They would need to know Passport.js session auth and how to build AES-256-GCM encrypted cookie auth from scratch. They would need to know Redux's dispatch-and-connect pattern and how to replace it with React Context providers and module-level refs. They would need to do all of this across 260 files without a single automated test to catch regressions 

), understanding intent, and verifying by hand. Finding a single human who holds all of this knowledge simultaneously, and who could execute on it in under a week without introducing regressions into a live production database, that person does not exist. 

. It is exactly the kind of problem that a machine with a large enough context window and deep enough code comprehension was built to solve.

Reading Supply is alive again. The migration is complete. And for the first time in 6 years, I can open my own codebase without dread. Sheesh.

What follows is the full accounting of what changed. If you're considering a migration of this scale, here you go.

The v0.3.3 codebase was a Next.js 9 application with a custom Express server, Passport.js authentication, Redux state management, Babel transpilation, and Slate 0.45 (Immutable.js-based) as the rich text editor. None of these survived the migration.

Every one of these changes was breaking. There was no upgrade path. We had only a full rewrite of the integration layer while keeping the business logic and database schema intact.

Breaking Changes: Next.js 9 to Next.js 16

Breaking Changes: Slate 0.45.0 to Slate 0.123.0

This was the hardest part of the entire migration. The data model changed completely:

The migration layer handles every content type in the production database: posts, comments (

), bios, signatures, archive descriptions, feedback responses, annotations, and encrypted content. Encrypted fields are decrypted, migrated, re-encrypted, and persisted. 

The migration solution we came up with is idempotent, it detects old format, skips new format, handles nulls, handles raw strings, handles already-migrated content. It runs at six integration points in the database query layer. Once content is saved in 0.123 format, there is no rollback without a full database backup. I think this was the best solution given I wanted everyone to come back to Reading Supply on their own timeline. We are supposed to keep your posts around forever.

Fifteen old plugin files were rewritten into eight new plugins

Over 300 files, 42,000 lines of code, 21,000 new lines, and 15 plugins later. Reading Supply is back after being held back by Slate 0.45.0

Reading Supply 0.4.0

 and I am so happy to finally announce this milestone. If you are curious about older updates you can view 

Aside from addressing editor issues, addressing markdown issues, and improving error messaging, there has been a bunch of memorable feature work that make this milestone memorable.

A lot of people wonder if I can read your posts on Reading Supply. 

I can not read your posts from the database, I can only read them from 

 and if and only if you have allowed it from 


For this milestone, I have completed some encryption work that ensures this guarantee further. It feels really great.

You can grant a specific user access to your post with a username.


Now regardless of your permissions settings, you can give a specific user on Reading Supply the ability to view your post. In the future this will expand into more things.

Previously a really annoying bug, today when a user exports a PDF on Reading Supply, any images that are included on the post will be included in the PDF version. 

Now you can keep your posts in PDF format offline!

Now you can view your library, write a post, or check your inbox from your phone. My sincere apologies to 

There are bugs with the writing experience on mobile. I'll be working hard to address that over the next couple of months.

 if you are a paying user. I'm excited about giving people a way to manage long form writing outside of Reading Supply's client.

You can view any Archive's RSS feed by appending 

 to the end of the URL. In the example below we are referencing 

This only works for archives and won't work for user profiles, since user profiles don't have a concept of a timeline.

There is still a ton to do before we ship the beta of Reading Supply:

Interested in trying out Reading Supply? 

All you have to do is ask a member of Reading Supply for an access key. If that feels like too much, you can direct message our 

. Once you get an access key you will be able to create an account and give others access.

Stay tuned for the next update, happy new year! 🎉

Learn more about our fifth alpha milestone.

Alpha Update 0.3.2

 Today I'm excited to share a new update about the 

. If you are curious about older updates, check out the development logs archive: 

.

The intent of the Tree Explorer is to give everyone a way to discover posts, archives and writers on Reading Supply. Unlike other popular social networks, 

the connections are made manually by the community

. An algorithm is not controlling or directing what you see, 



Originally, I shared the early concept in an 

. After some work, this is what I ended up with:



The implementation is simple, here is how it works:


As you navigate to nodes, any node with a post, archive, or writer mention will appear. Even with our small community, the tree can get pretty big!


If a writer mentions a post, archive, or another writer in their biography, the Tree Explorer will show those relationships too. The same effect applies to archives as well.




Navigating all of these nodes should feel fun. When I was in high school. I would get together with my friends in the library and play the 

. Players would start on the same random article on Wikipedia, and each player must navigate to another pre selected target article. As Reading Supply grows I hope there is enough content to be able to have a similar experience.


The Tree Explorer solves practical problems too. It should be easy to find the origin of any writing such as a white paper. It should also be easy to discover the next thing to read within the context of what you are reading. Therefore I made navigating to the 

 page convenient. Every mention of a writer, archive, or post shows a link to the Tree Explorer. Every post and user profile also has a link.


Currently, the Tree Explorer defaults to an archive named 

. Anyone is welcome to join the archive and publish to it.

It is important to get feedback early. So I shipped this release knowing there are problems.

If you discover something that bothers you, please share! I want to know anything and everything! 🙏

I will be addressing as many as I can. Here are a few reported by the community:


Interested in trying out Reading Supply? All you have to do is ask a member of Reading Supply for an access key. If that feels like too much, you can direct message our 

. Once you get an access key you will be able to create an account and give others access.

Till next time 👋

Learn more about the release with our Tree Explorer prototype

Alpha Update 0.3.1

I am excited to share this update with everyone! A long time has passed since I wrote about our progress. 

 was the last published post and that post was published half a year ago. If you are curious about older updates, take a look at the 

I am happy people are writing. I can't read most of the posts. However, I can see when posts are published 

if the public is allowed access through permissions


These observations make me optimistic about the future. 🎉

 will begin to thrive when people write about their learnings.  Here are some new additions to help us get there:


Connecting posts, archives, and users together

, you may notice some new inline elements:


 symbol. You can also reference an archive by using the 

 I want to show the connection between posts, archives, and users

. This vision will become clear once I deploy the 



The lack of auto completion and search is by design. I understand this choice is counter intuitive because popular products usually provide this functionality. However, I want people to reference content they were directly influenced by.

Give this feature a try in your next post ✍

 posts in your library threshold (as discovered by 

 ), you may desire a way to search all the paragraphs and posts you have written. 

This feature has helped me go through old knowledge easily. I felt it was a way I could rediscover my attempts to define terms and concepts. I also found myself feeling refreshed on ideas I use to believe, and I enjoyed the option to make revisions wherever I felt necessary.

After shipping search, I wanted the posts in the library to feel simple and informative.


For users who don't spend all day on a machine, a human readable sentence can provide more clarity than symbols and adjectives. Let me know if you feel the same way!

During a discussion, people may ask you for your references. They will want to know how you do your work (process) and how you formulated your beliefs. We are making the bet that a personal library that is easy for you to query will be essential for an in-depth conversation. Maybe we could take this further by providing a way to allow users to give consent for merging their libraries.

What if it felt like merging minds? 🧠


In the past, our editor had buttons that did nothing. That had to change.

A lot of embarrassing issues were in production. But the good news is that a lot of bugs have been fixed since 

 was released. I no longer have to feel that awkward 

“you-know-some-big-things-are-broken”

Please let me know if you find anything that bothers you!

I added a way for paying members to use custom domains:

Setup should only takes a few minutes. The amount of time may vary because the process involves a couple of round trips between you and your registrar. We will handle SSL certification for you. 

I am really excited people are using 

 profiles to take the place of their personal website! Here is the first cohort:

I will invest time improving the security and infrastructure. I am happy with the state of things, but I know there are some bad actors on the internet and you need to be prepared for everything.

In the near term, I have some brand and identity ideas I'd like to try out! Please don't hesitate to share your opinions. Help me imagine the digital version of the 

 will be a fun feature to build. Users will be able to navigate all the connections between their posts, archives, discussions, and mentioned users. I think the ideal experience will be just as fun as 

 or any compelling text based adventure. I hope it becomes another reason why people hang out here.

Thanks again for your time, it means the world 🌎.

Joining Reading Supply is free if you're willing to put in a little effort. All you have to do is ask a member of 

 for an access key. If that feels like too much, you can direct message our 

. Once you get an access key you will be able to create an account and give anyone free access too.

But if you are feeling anti-social you can always visit 

Learn more about the third alpha milestone.

Alpha Update 0.3.0

, you can read about previous updates here: 

Up until this update, Reading Supply has been focused on the single player writing experience. However organic and meaningful conversations are a part of why Reading Supply is important. Unfortunately, there were some big problems:

So for new years day, activity feeds and improvements to inbox notifications were deployed to help solve this problem.


When you join Reading Supply or sign in to an existing account, you're greeted with this new screen.

On this page, any messages, feedback, or annotations you receive or send will appear. In addition, you will get updates for published posts from archives you are a part of.

Here is an imaginary scenario for you:

If you look in the top left, you'll see a red indicator for the number of new activity events.

After this update, I am confident there is reliable structure in place for keeping track of interactions with other writers and readers. For the short term I will focus on the editor and performance. I look forward to sharing what I have planned. :-)

If you have ideas, share them with me on 

Learn more about a small update before the third alpha milestone.

Alpha Update 0.2.8

Welcome back! You are reading a development update for 

For those of you who do not know, Reading Supply is a reading and writing tool for artists, thinkers, and creators. I am focused on the following:

Libraries are a way to organize your drafts and posts.


I want people to have more organizational freedom. You can drag and drop posts into different folders. If you drag a post above or below another post you can change the order, which will update the order on your profile.

Libraries creates space for more functionality in the future. People will be able to import their markdown, JSON or third-party posts. In addition, I want to introduce a post metrics dashboard, and a way to see deep link connections between posts.


Archives are a way people can share a timeline either publicly or privately.

Owners of an archive can invite anyone to join. Posts appear in order of the publishing date. In the settings, the owner can choose to make the archive page only viewable by members.

Archive pages are perfect for any group of people who want to collaborate. Eventually I will add customization and the capability to create a physical book with posts from your archive.

Writers have more control over who can view their drafts and published posts.

By default a draft is only visible to the author, and a published post can be viewed by anyone who has the link. You can publish a post into as many archives you are a member of.

People without access to your post will see this:

The posts on your profile are organized by folder.

Threads are no longer fully exposed at the bottom of a post.


Users will need to click into each thread to view the full discussion in the sidebar.

Annotations UI has been tweaked too:


The next major update will be focused on the editor.

If you have ideas please share them with me on 

. I love to entertain any idea or suggestion.

Thank you for your time! 🙏

Learn more about the second alpha milestone.

Alpha Update 0.2.7

It has been a couple of months since I started

 I am really excited to share the progress. This is a great opportunity to reflect on completed features and look at the big picture.

Are you interested in trying Reading Supply? Find me on 

I love the writing produced by artists, thinkers, creators, and thoughtful humans. I built Reading Supply because I believe everyone should have access to good tools for developing and sharing their specific knowledge. I think good processes are enhanced by tools that achieve perfect plainness and maximum usability.

Programming is simple when you have a clear direction, but early products need a compelling north star. To address this, I conducted some research and took notes. Here is a rough summary:

On top of that, there are many impressive tools and platforms that exist today:

With all things considered, it would be a fool's errand to try to create a thesis for how Reading Supply will become the best reading and writing experience on the Internet today. The most brilliant tools have been worked on for many years and have large teams learning and iterating on them. 

There is so much that I still need to learn, so I am going to focus on this basic system for the next alpha milestone 

 Always ensure these 7 claims are always true:

: 1+ days of the week spent away from implementation:

I'm confident that adhering to this system will lead somewhere. Since alpha 

 is feature complete, lets take a brief walkthrough of the product as it exists today.

This is what you see when you open your browser and start writing:

I picked commonly-seen "block" and "inline" styles from other products. But I have also added some other unique blocks to the available options such as a 

I added a live version to this post for fun:

I will admit, outside of this hat trick, the editor will probably feel a standard web  

WYSIWYG editor from 2018, but that is probably a good thing. 😎

Once the user has written a post and saved it, the post appears in the drafts section of the author's profile. When they publish their draft it will appear in the published section of their profile. 

. The hope is that users will treat their profile like a personal homepage and share it frequently.  🙏

Writers will want to get peer review on their drafts before they are published publicly. A lot of professional products have inline commenting in-between text blocks or messages that appear on the right side of the screen. In addition, the author may want to have a discussion so message threads must exist as well.

Reading Supply takes a bet that you'll want to drop a feedback pin instead. This behavior is stolen from design tools where collaborators drop pins on a specific part of the image they wish to provide feedback on. On our end, we can treat the whole long form post just like an image.


Here is an example of the reader being able to drop the pin anywhere:

Clicking on the post will turn the pin green. Afterwards you can leave messages for the author by clicking on the pin.

Anyone who is logged in and has access to your draft's unique URL will be able to provide feedback. Feedback is viewable on both web and mobile, and none of the feedback will appear on your published post even if you don't remove each one manually. 

To make things interesting, I am also experimenting with removing vanity metrics from an author's published post. This is controversial because vanity metrics can act as a form of encouragement. It feels good to know who agrees with your stance and it is a low friction task for a reader to 


But without vanity metrics, there is potential for discussions to be redesigned with a clean slate. Currently, when an author publishes their post, other authenticated readers can start a message thread at the bottom and annotate any section of text. 


I also can't help but feel that we're exploring Marc Andreessen's idea:

I often wonder how the Internet would have turned out differently if users had been able to annotate everything – to add new layers of knowledge to all knowledge, on and on, ad infinitum.

Here is what a discussion thread looks like to the reader:


Even though rich text features are enabled for inline styles, this resembles a typical blog post on the internet.

Learn more about the first alpha milestone.