Coding

Background Okay, so this post is a follow up to a few different posts, but most notably The Token Saving Cake is a Lie. Before reading this one, I highly recommend checking that out first. Go ahead. I’ll wait. Back? Good. Let’s continue. Headroom So I said that the solution is probably something like Lynkr, but before building the integration, I discovered Headroom (45k stars, Python, its own trained compression model, ships RTK as a component). It does everything the Lynkr plan described, supposedly better. Set ANTHROPIC_BASE_URL=http://localhost:8787, run headroom wrap claude, done. ...

Previous posts for context: The Future of AI Coding is Directing and C.A.S.T Progress: First Blood Last time I had one event flowing from a ticket board to a NATS topic. That was the plumbing. This time the plumbing has something running through it: a ticket goes in one end as a one-line wish, and a reviewed, merged, cleaned-up branch comes out the other. No human types a line of code in between. ...

Previous post for context: The Future of AI Coding is Directing. Let Me Introduce You to Your C.A.S.T. Also, this banner was generated by ChatGPT and whatever model they are using at the moment, and I’m kinda impress, NGL. When I wrote about C.A.S.T last month, the design was solid but unproven. I had a SPEC, ADRs, hire/fire scripts, and a lot of conviction. What I didn’t have was a single event flowing from a ticket board to a NATS topic to a waiting agent. ...

Introduction I came across caveman-compression — a tool that compresses text into a token-efficient “caveman” format before sending it to an LLM. The pitch is simple: It can cut 30-65% of tokens depending on the content, which at Claude Opus prices adds up fast. I immediately wanted to wire it into Claude Code so compression would happen automatically on every prompt. The obvious hook was UserPromptSubmit. ...

Previous post for context: Two AI Agents Walk Into a Codebase If you look at social media, everyone will tell you that most people are still using AI like a very fast keyboard. You write what you want, it writes the code, you review it, you move on. Faster, sure. But you’re still in the driver’s seat for every line, every file, every decision. They will also tell you that the future is not that. The future is swarms of AI agents doing the work of 100s of people. And I mostly agree. The shift that’s actually happening — and I say this as someone who’s spent the last month living inside it — is from writing code to directing agents. ...

I never posted why I paused development on So This Is How I Die. If you’ve been following the devlog, you’ll know the story up to a point. The slow progress, the balancing act with client work, the occasional “I’m having a hard time” entry that I posted because solo dev accountability sometimes means confessing that nothing is on fire, you’re just tired. What you won’t find is a post that says this is where I stopped, and here’s why. ...

That’s the punchline and the problem. I’ve run into these two similar scenarios which have had me stumped and annoyed. In one, I’m running Claude Code in both GoLand and PyCharm, trying to integrate one project’s API with another project’s GRPC service. In the other, I have a Scala library that needs updates, and then multiple services that depend on that library. The hiccup: each instance is completely unaware of the other’s existence. ...

🧠 Something I’ve been thinking about a lot after doing these AI coding experiments… I get why non-devs enjoy this so much. Creating something cool gives you a dopamine hit. Solving a problem does it too. Just like how YT Shorts, IG Reels, TikTok and even LinkedIn these days give you these hits through scrolling, these AI tools do it by “doing, showing, doing, showing”. ...

I’ve been playing around with Cursor and Windsurf, because I keep seeing on social media how great these tools are and how easy it is to build things. And you know what? It’s…. ummm… better than before? Okay, look, it’s impressive. If I compare it so some earlier versions of the IDEs and other tools like Bolt.new, sure, it’s improved. But sheesh, does it take poking and prodding and coddling to get somewhere. Maybe I’m using it wrong, I don’t know, but I wish I could share the transcripts of my chats with these tools so you’ll understand… (can I export those? am I missing a feature somewhere?) ...

🎯 This morning’s YouTube recommendation: someone coding purely by voice using Cursor agents and Windows dictation tools. My first thought? Not “wow, how cool!” but “seriously, is typing that hard now!?” 🙄 My second thought: “Good luck remote debugging a Linux server at 1am to fix what your AI buddy broke…” First we outsource thinking to AI, now the typing… it’s like nobody wants to actually do this job anymore! 🤦 But here’s the thing - after my initial skepticism, I realized this IS actually pretty cool. Just not for the reasons filling up your social media feed. ...