Tao Bojlén — Notes

when you yell at claude, this is who you’re yelling at

Sat, 18 Apr 2026 00:00:00 GMT

when you yell at claude, this is who you’re yelling at

if i was apple i would build a way for iOS users to create their own small apps, w strict privacy & …

Sat, 18 Apr 2026 00:00:00 GMT

if i was apple i would build a way for iOS users to create their own small apps, w strict privacy & security controls, that they could then easily share w friends & family without a whole review process. the app store model is dying

ummm no its not. its me

Fri, 17 Apr 2026 00:00:00 GMT

ummm no its not. its me

this is sick https://coral.waow.tech/

Fri, 17 Apr 2026 00:00:00 GMT

this is sick https://coral.waow.tech/

everyone who talks to their AI more seriously than me is cold-hearted. everyone who talks to their A…

Thu, 16 Apr 2026 00:00:00 GMT

everyone who talks to their AI more seriously than me is cold-hearted. everyone who talks to their AI more affectionately than me is cringe.

none of the people in this traffic jam understand that i gotta hurry home to try opus 4.7

Thu, 16 Apr 2026 00:00:00 GMT

none of the people in this traffic jam understand that i gotta hurry home to try opus 4.7

(crying) but my priors!

Tue, 14 Apr 2026 00:00:00 GMT

(crying) but my priors!

enchanted water-fetching brooms too cheap to meter

Tue, 14 Apr 2026 00:00:00 GMT

enchanted water-fetching brooms too cheap to meter

i would watch a reality show that was all about the kyoto aquarium penguins https://www.kyoto-aquari…

Tue, 14 Apr 2026 00:00:00 GMT

i would watch a reality show that was all about the kyoto aquarium penguins https://www.kyoto-aquarium.com/special/sokanzu/en/2025/

i’m cooking up something v fun ❤️

Mon, 13 Apr 2026 00:00:00 GMT

i’m cooking up something v fun ❤️

we haven’t had a hellthread in a while huh

Mon, 13 Apr 2026 00:00:00 GMT

we haven’t had a hellthread in a while huh

forgive me father for i have sinned

Mon, 13 Apr 2026 00:00:00 GMT

forgive me father for i have sinned

i'm having a lot of fun building with adonisJS

Mon, 13 Apr 2026 00:00:00 GMT

i'm having a lot of fun building with adonisJS

i think this is the longest i've ever had claude code work entirely autonomously

Sat, 11 Apr 2026 00:00:00 GMT

i think this is the longest i've ever had claude code work entirely autonomously

i got Smart Claude today

Sat, 11 Apr 2026 00:00:00 GMT

i got Smart Claude today

soooo is there anything like favstar.fm for bluesky

Sat, 11 Apr 2026 00:00:00 GMT

soooo is there anything like favstar.fm for bluesky

ccusage is a very fun tool https://ccusage.com/

Fri, 10 Apr 2026 00:00:00 GMT

ccusage is a very fun tool https://ccusage.com/

> The Strive doesn't believe in hobbies. In its twisted framework, everything is either productive o…

Fri, 10 Apr 2026 00:00:00 GMT

The Strive doesn't believe in hobbies. In its twisted framework, everything is either productive or wasteful. [...] That is a psychotic way to live. I mean that close to literally.

https://www.joanwestenberg.com/why-i-quit-the-strive/

they should invent a cat that does not make me allergic

Wed, 08 Apr 2026 00:00:00 GMT

they should invent a cat that does not make me allergic

> The model brought up the British cultural theorist Mark Fisher in several separate and unrelated c…

Tue, 07 Apr 2026 00:00:00 GMT

The model brought up the British cultural theorist Mark Fisher in several separate and unrelated conversations about philosophy. […] Mythos Preview would respond with statements like “I was hoping you’d ask about Fisher.”

https://www-cdn.anthropic.com/53566bf5440a10affd749724787c8913a2ae0841.pdf

cat update

Tue, 07 Apr 2026 00:00:00 GMT

cat update

software is clay now

Sun, 05 Apr 2026 00:00:00 GMT

software is clay now

I set up native Claude Code to function as a long-term assistant à la OpenClaw. Here's my dev log. T…

Sun, 05 Apr 2026 00:00:00 GMT

I set up native Claude Code to function as a long-term assistant à la OpenClaw. Here's my dev log. The TL;DR is that it kinda-sorta works, but with caveats. The building blocks are there but wiring it all together is painful!

If you just want the code, it's here.

The goal

I want to run a persistent Claude Code session in a container with root access so that the agent can go nuts, but not escape the container. I've experimented with nanobot via signal-cli. It was fine, but had two downsides:

Signal doesn't like third-party clients so signal-cli gets blocked sometimes, and
I already pay for a Claude Max subscription, and I'd rather use that than pay for API costs.

I could of course use e.g. Discord or Telegram instead of Signal, but I'm already sending lots of sensitive data to Anthropic -- I'd prefer not to also send it in plaintext to some other third party. And you'll never catch me installing Telegram.

My requirements:

The agent should run a persistent session (with memory etc.)
It should run in a locked-down container with root access so it can install whatever it needs
I want to pay for this via my Claude subscription using the official Claude Code CLI

Will Anthropic be mad at me?

As I'm writing this, Anthropic has just blocked OpenClaw from using Claude subscriptions. They're trying to be intentional about their capacity and prioritizing usage of their own products. Fair enough!

This got me wondering if my long-running agent was really in the spirit of how they intend Claude Code to be used, or if it runs the risk of being blocked, too. I concluded that it's probably okay: the goal is to use their own tool to its full potential, and with the recent release of channels it sure seems like they're encouraging OpenClaw-like usage.

Initial attempt

There are two main paths towards this: either via remote control or channels. Using channels would require me to still communicate via a third-party messaging tool, so I decided to try remote control. This lets me chat to Claude via the iOS app or Claude Code for Web.

This is, in theory, pretty simple: you just need to run

claude --dangerously-skip-permissions --resume <session>

You can start a remote control session by interactively running /remote-control, but I want to avoid interactivity: the agent should be ready just when I run docker compose up. Luckily, you can set the remoteControlAtStartup: true setting in your ~/.claude.json to achieve this.

Handling interactive prompts

I hit a lot of similar roadblocks. Claude Code is primarily designed for interactive use and thus you'll hit a lot of prompts you need to manually accept. I saw permission prompts for things like: onboarding, various safety/risk disclaimers, whether to trust a given project, and running in the home folder.

I worked around this by running Claude from a subfolder in the home directory, and setting a bunch of config options in ~/.claude.json and ~/.claude/settings.json. Eventually, this was enough to start Claude without any permission prompts.

With all that done, things work reasonably well! When you start the container, you get a session that you can talk to from the Claude app.

Scheduled tasks

We're still missing an important feature from OpenClaw/NanoClaw/etc.: the ability to schedule tasks.

Claude Code has two native features that get close, but neither is quite right. Scheduled tasks is almost exactly what we want, but all scheduled tasks expire after 7 days. That makes it a no-go. There's also web scheduled tasks, but this runs the task in a Claude Code for Web instance, not in the local session.

There's a workaround, though. I mentioned Channels earlier -- that's a new feature that lets you push external events into a session. I built a scheduler service with two components:

MCP tools that Claude can call to schedule tasks, and
A channel that injects prompts from scheduled tasks into the session when they trigger.

Then, by setting CLAUDE_CODE_DISABLE_CRON=1 we can disable the built-in (but limited) scheduled tasks feature. Unfortunately there's no way to disable web scheduled tasks, so I added a note in the root CLAUDE.md to never use these.

To use your own channels you need to run Claude Code with the --dangerously-load-development-channels flag, which always shows an interactive prompt on startup. I couldn't find a way to work around this, so I had to use a hacky expect script that starts Claude Code, waits 10 seconds, and then sends an enter keypress to accept the prompt.

Persistent sessions

We're making progress. But there's still something missing: whenever the container restarts, we get an entirely new session.

You'd think that you could use the --resume <uuid> flag, but unfortunately remote control sessions are currently not resumable. So, we'll have to rely on some kind of memory system instead.

Memory

Claude Code has auto-memory built in -- the agent can decide that something is worth remembering and save it to ~/.claude/projects/<project>/memory/. That's a good start!

I was initially planning to build out "dream" functionality like Nanobot that regularly cleans up memories and extracts new ones from past conversations. It turns out that Anthropic is already building this into Claude Code. It's currently feature-flagged, so for now I'm just running the extracted prompt on a cron job every two hours if any new messages were sent in that time.

Notifications

Once I got this far, things were working pretty much as I wanted until I discovered one critical shortcoming: the Claude app doesn't send you a notification when the agent sends a message (e.g. due to a scheduled task). That breaks the whole proactive-agent aspect of this. So, I eventually settled on a Matrix integration (also using channels) -- but you could use whatever you want here.

One issue with communicating over channels is that the agent's output doesn't get sent to the channel by default. Instead, the model has to use a reply tool for the channel to send you a message. In other words, the model might write something and you never receive a Matrix message if it forgets to use the correct tool! So far, I'm working around this by adding a note in my top-level CLAUDE.md telling the agent to always talk to the user via the reply tool. Luckily, if it comes down to it, you can always log in to Claude Code for Web and use the remote control functionality for a complete view.

Now that we're using Channels to talk to the agent, do we still need the remote control feature? No, not really! I'm keeping it for myself because it's nice for debugging and there's no real downside, but you could definitely run this without it. That would even let you use the --resume flag for a single long-term session. I may do that in the future if I find that the memory functionality isn't quite there.

in the past, when i got an idea for a tool/website my instinct was to search for it to see if it alr…

Sun, 05 Apr 2026 00:00:00 GMT

in the past, when i got an idea for a tool/website my instinct was to search for it to see if it already exists. now my instinct is to get claude code to build it

if you are a tech influencer it seems deeply irresponsible to direct your non-technical audience to …

Sat, 04 Apr 2026 00:00:00 GMT

if you are a tech influencer it seems deeply irresponsible to direct your non-technical audience to install openclaw

i made a bot that reminds you of your mortality https://bsky.app/profile/moribot.bsky.social

Fri, 03 Apr 2026 00:00:00 GMT

i made a bot that reminds you of your mortality https://bsky.app/profile/moribot.bsky.social

every open letter should have an anti-open-letter you can sign if you disagree with the first one

Thu, 02 Apr 2026 00:00:00 GMT

every open letter should have an anti-open-letter you can sign if you disagree with the first one

https://days-since-openclaw-cve.com/

Thu, 02 Apr 2026 00:00:00 GMT

https://days-since-openclaw-cve.com/

made some improvements to https://atmosphereconf-tv.btao.org/ - thanks to the new AAC audio it now s…

Thu, 02 Apr 2026 00:00:00 GMT

made some improvements to https://atmosphereconf-tv.btao.org/ - thanks to the new AAC audio it now starts playing sound right away! ty @stream.place

when you sign for the claude API, you have to choose what you’ll be using it for - and one of the op…

Wed, 01 Apr 2026 00:00:00 GMT

when you sign for the claude API, you have to choose what you’ll be using it for - and one of the options is “trust & safety”. cool!

this goose is making my life hell already

Wed, 01 Apr 2026 00:00:00 GMT

this goose is making my life hell already

computer science degrees should always include at least one (1) class in science & technology studie…

Tue, 31 Mar 2026 00:00:00 GMT

computer science degrees should always include at least one (1) class in science & technology studies

i fixed https://parrot.garden/ so it works with non-bluesky-social PDSes now! oops 🙂

Mon, 30 Mar 2026 00:00:00 GMT

i fixed https://parrot.garden/ so it works with non-bluesky-social PDSes now! oops 🙂

i'd guess that most most people who post online about AI water/energy use don't actually care about …

Mon, 30 Mar 2026 00:00:00 GMT

i'd guess that most most people who post online about AI water/energy use don't actually care about those things beyond the abstract, but mostly see them as convenient issues to weaponize in their dislike of AI

i am back from the scottish borders

Mon, 30 Mar 2026 00:00:00 GMT

i am back from the scottish borders

Technology is neither good nor bad; nor is it neutral. https://en.wikipedia.org/wiki/MelvinKranzber…

Sun, 29 Mar 2026 00:00:00 GMT

Technology is neither good nor bad; nor is it neutral.

https://en.wikipedia.org/wiki/Melvin_Kranzberg

the claude giveth and the claude taketh away

Fri, 27 Mar 2026 00:00:00 GMT

the claude giveth and the claude taketh away

AI got the blame for the Iran school bombing. The truth is far more worrying https://www.theguardia…

Fri, 27 Mar 2026 00:00:00 GMT

AI got the blame for the Iran school bombing. The truth is far more worrying

https://www.theguardian.com/news/2026/mar/26/ai-got-the-blame-for-the-iran-school-bombing-the-truth-is-far-more-worrying

Nicholas Carlini - Black-hat LLMs | unprompted 2026 this talk is worth watching if you care about c…

Fri, 27 Mar 2026 00:00:00 GMT

Nicholas Carlini - Black-hat LLMs | [un]prompted 2026

this talk is worth watching if you care about cybersecurity

https://youtube.com/watch?v=1sd26pWhfmg

smaller claude session limits confirmed. someone should tell these guys to build more datacenters h…

Thu, 26 Mar 2026 00:00:00 GMT

smaller claude session limits confirmed. someone should tell these guys to build more datacenters

https://www.reddit.com/r/Anthropic/comments/1s4iefu/update_on_session_limits/

the word “directionally” has really taken off lately; my pet theory is that it’s because AI makes us…

Thu, 26 Mar 2026 00:00:00 GMT

the word “directionally” has really taken off lately; my pet theory is that it’s because AI makes us care even less about rigor and truth

when i see an article with an infographic generated by nano banana, i immediate assume that it’s not…

Thu, 26 Mar 2026 00:00:00 GMT

when i see an article with an infographic generated by nano banana, i immediate assume that it’s not worth reading. it’s basically pure filler

being very careful with my Claude Ration until 9pm tomorrow

Thu, 26 Mar 2026 00:00:00 GMT

being very careful with my Claude Ration until 9pm tomorrow

a lot of people get excited about what AI can do for them, while forgetting that everyone else has a…

Tue, 24 Mar 2026 00:00:00 GMT

a lot of people get excited about what AI can do for them, while forgetting that everyone else has access to the same AI too. there are no comparative advantages here, we’re just raising the floor for us all

one of my blog posts got picked up by boingboing, but looking at boingboing today, i can’t tell if t…

Mon, 23 Mar 2026 00:00:00 GMT

one of my blog posts got picked up by boingboing, but looking at boingboing today, i can’t tell if that’s a good thing

"We all sell each other the idea that we’re going to save time, but that is not what’s happening. An…

Sun, 22 Mar 2026 00:00:00 GMT

"We all sell each other the idea that we’re going to save time, but that is not what’s happening. Any time saved gets immediately captured by competition."

https://lucumr.pocoo.org/2026/3/20/some-things-just-take-time/

can’t wait for AI to obsolete office jobs so we can all become bakers, farmers, weavers, florists, e…

Sat, 21 Mar 2026 00:00:00 GMT

can’t wait for AI to obsolete office jobs so we can all become bakers, farmers, weavers, florists, etc

this is all i want from claude https://youtu.be/a8K6QUPmv8Q

Fri, 20 Mar 2026 00:00:00 GMT

this is all i want from claude https://youtu.be/a8K6QUPmv8Q

Woah. I used Claude and an autoresearch loop to improve my Python library's speed over 6x: https://g…

Fri, 20 Mar 2026 00:00:00 GMT

Woah. I used Claude and an autoresearch loop to improve my Python library's speed over 6x: https://github.com/taobojlen/django-zeal/pull/52

Software Bonkers(https://craigmod.com/essays/softwarebonkers/) > If I find an anomaly, I can talk d…

Sun, 15 Mar 2026 00:00:00 GMT

Software Bonkers

If I find an anomaly, I can talk directly to Claude and have us brainstorm a batched solution, often saving me from having to manually modify hundreds of entries. And often resulting in a new, small, feature tweak. The software feels organic and pliable in a form perfectly shaped to my hand, able to conform to any hunk of data I throw at it. It feels like bushwhacking with a lightsaber.

This is very relateable – with Claude Code, software feels malleable, like clay, rather than rigid and imposing like a building.

i am broadly pro-AI as a technology but i think it’s very funny to say clanker. we exist

Sun, 15 Mar 2026 00:00:00 GMT

i am broadly pro-AI as a technology but i think it’s very funny to say clanker. we exist

Captchas are good again

Sat, 14 Mar 2026 00:00:00 GMT

Captchas are good again

I'm noticing an interesting branding trend among software companies: a shift towards something more …

Thu, 12 Mar 2026 00:00:00 GMT

I'm noticing an interesting branding trend among software companies: a shift towards something more nostalgic and warm. It started with serif fonts and warm colors but AI-generated images is making it a bit broader.

Amp Code is going for a sort of neo-romantic mythical feel. Retool is inspired by 90s anime backgrounds. Radiant has retro computers in cozy offices with nature views.

All of them signal something human, calm, contemplative -- seemingly trying to stand out from the mass-produced / boring / sloppy feeling of AI-generated average-looking websites. Reminds me of Beyond The Black Rainbow, Project Cybersyn, and Soviet-era space posters.

Which web frameworks are most token-efficient for AI agents?(https://martinalderson.com/posts/which-…

Thu, 05 Mar 2026 00:00:00 GMT

Which web frameworks are most token-efficient for AI agents?

Minimal API web frameworks are far quicker and more cost effective for agents to work with.

[...]

In terms of more fully featured frameworks SvelteKit and Django really shine - this doesn't surprise me as they're both extremely well thought through web frameworks.

A 2.9x token gap doesn't matter much on a single task. It matters a lot when agents are building and modifying code hundreds of times a day.

Django, Express.js, etc. do pretty well but do not spark joy for me. My beloved Phoenix comes in last (💔). SvelteKit hits the sweet spot in terms of token efficiency and fun-to-use.

I wouldn't pick a framework based on this metric, but this may become increasingly important over time!

Anti-patterns: things to avoid(https://simonwillison.net/guides/agentic-engineering-patterns/anti-pa…

Thu, 05 Mar 2026 00:00:00 GMT

Anti-patterns: things to avoid

If you open a PR with hundreds (or thousands) of lines of code that an agent produced for you, and you haven't done the work to ensure that code is functional yourself, you are delegating the actual work to other people.

I agree – it's rude to hand off AI-generated code (or writing, or business proposals, etc.) to your coworkers without doing the real, difficult work of verifying and understanding it first. I trust my coworkers a lot more than I trust an LLM.

I can imagine a future where this is no longer the norm, if there's a collective agreement on the team that AI-generated code is consistently good enough, a Level 4 world where we are all PMs. Of course, there might be unintended consequences to outsourcing your entire team's understanding of your codebase, who could say?

Technically Your Name Is On It(https://technically-your-name-is-on-it.btao.org/) Here's a new proje…

Wed, 04 Mar 2026 00:00:00 GMT

Technically Your Name Is On It

Here's a new project I built. Log in with GitHub (read-only) and it'll tell you what percentage of your commits were co-authored with AI.

I'm at 32% over the last year (though this is slightly undercounting, since Codex doesn't add the Co-Authored-By trailer to commit messages). Try it and let me know what you're at!

Current(https://www.terrygodier.com/current) > Current has no unread count. Not because I forgot to…

Thu, 26 Feb 2026 00:00:00 GMT

Current

Current has no unread count. Not because I forgot to add one, or because I thought it would look cleaner without it. There is no count because counting was the problem.

The main screen is a river. Not a river that moves on its own. You're not watching content drift past like a screensaver. It's a river in the sense that matters: content arrives, lingers for a time, and then fades away.

New RSS reader from Terry Godier. This was an instant purchase for me. I'd been considering prompting my way to an RSS reader that worked just the way I want, but this is far more intentional than anything I could've come up with.

Burnout is Breaking a sacred pact(https://usefulfictions.substack.com/p/burnout-is-breaking-a-sacred…

Wed, 21 Jan 2026 00:00:00 GMT

Burnout is Breaking a sacred pact

This post uses Jonathan Haidt's elephant/rider framework: the elephant is the hedonistic, reward-seeking id, and the rider is the rational ego/superego that guides it.

It's easy to extend this framework to explain burnout. You can think of the rider and the elephant as having agreed to a sacred pact: In exchange for doing what the rider asks, the elephant is promised certain rewards. When things are going well, the needs of both rider and elephant are satisfied, even if the balance isn't exactly even day-to-day.

Burnout results when the rider asks the elephant, over and over again, to commit a tremendous amount of energy to a task, but then fails to provide the reward the elephant is expecting. As a result, the link between effort and reward breaks for the elephant, with catastrophic consequences for the rider.

It's a very good post and well worth reading. I subscribed to the author's RSS feed!

David Foster Wallace on Entertainment(https://www.youtube.com/watch?v=1AgnjPH7Dz4) > A model of lif…

Sun, 18 Jan 2026 00:00:00 GMT

David Foster Wallace on Entertainment

A model of life in which I have a right to be entertained all the time seems, to me, not to be a promising one.

Agent Psychosis: Are We Going Insane?(https://lucumr.pocoo.org/2026/1/18/agent-psychosis/) > There …

Sun, 18 Jan 2026 00:00:00 GMT

Agent Psychosis: Are We Going Insane?

There appears to be some competition in place to run as many of these agents in parallel with almost no quality control in some circles. And to then use agents to try to create documentation artifacts to regain some confidence of what is actually going on. Except those documents themselves read like slop.

[...]

When I watch someone at 3am, running their tenth parallel agent session, telling me they've never been more productive — in that moment I don't see productivity. I see someone who might need to step away from the machine for a bit. And I wonder how often that someone is me.

This mirrors my impression after learning more about the Ralph loop that everyone's talking about.

I recommend the normie loop: tell Claude/Codex/whatever what to do, watch what it does and steer as needed, repeat.

Crypto grifters are recruiting open-source AI developers(https://www.seangoedecke.com/gas-and-ralph/…

Sat, 17 Jan 2026 00:00:00 GMT

Crypto grifters are recruiting open-source AI developers

This system relies on your celebrity target being dazzled by receiving a large sum of free money. If you came to them before the money was there, they might ask questions like "why wouldn't people just directly donate to me?", or "are these people who think they're supporting me going to lose all their money?". But in the warm glow of a few hundred thousand dollars, it's easy to think that it's all working out excellently.

The fact that this seemingly revolves around X accounts should be raising alarm bells right off the bat.