# bitcoin wallet — RSS Amplifier Recent posts from the 2 feeds in the RSS Amplifier directory that cover bitcoin wallet. Page: Feed: --- ## [I built a Greek pottery roguelite with Claude Code in a weekend](https://josh.miami/posts/sisyphus-greek-pottery-roguelite) _2026-08-16 · josh.miami_ Project Sisyphus went from an empty folder to a playable roguelite on my iPhone. The best design decision was noticing that programmer art and ancient vase painting are the same thing. ## [Anthropic's Mythos card reads like a containment memo](https://josh.miami/posts/mythos-containment-memo) _2026-04-08 · josh.miami_ Anthropic built a stronger model and decided not to broadly ship it. That alone separates this from the usual frontier-model promo cycle. Skynet is a stupid frame for it. What's actually in the card is duller and a lot more serious. ## [I built a starter template because I kept rebuilding the same thing](https://josh.miami/posts/monolith-industries) _2026-03-21 · josh.miami_ Monolith Industries is a Next.js starter kit with auth, Supabase, Sentry, background jobs, and every tooling decision already made. It came from a real product. That's the whole point. ## [I built a Pokemon game inside Claude Code's statusline](https://josh.miami/posts/statusmon-day-one) _2026-03-14 · josh.miami_ What started as 'can I put a Pokemon in the status bar' turned into a full companion system with ANSI sprites, evolution chains, and a Pokedex. All in one session. ## [TokenGolf hit v1.0 because the two-mode system was the wrong abstraction](https://josh.miami/posts/token-golf-v1) _2026-03-13 · josh.miami_ The wizard asked you to commit a budget before you knew what the session needed. Par replaced that question with a better one: how efficient were you given what actually happened? ## [What I shipped today in TokenGolf v0.4.0](https://josh.miami/posts/token-golf-v040) _2026-03-09 · josh.miami_ Expansion and hardening at the same time. More achievements, more hooks, more structure, and the release infrastructure to keep the rules from drifting. ## [Sunday was when TokenGolf stopped feeling bolted on](https://josh.miami/posts/token-golf-sunday) _2026-03-08 · josh.miami_ By the end of Saturday, the score was believable. Sunday became the day of friction removal and deeper integration. ## [The repo looked done until I tried to run it](https://josh.miami/posts/token-golf-saturday) _2026-03-07 · josh.miami_ Saturday was where TokenGolf stopped being a neat concept and started becoming a real tool. The first milestone had nothing to do with the game. ## [The Half-Second Buffer](https://josh.miami/posts/the-half-second-buffer) _2026-03-06 · josh.miami_ You're doing your best work. You get comfortable. Then you catch a hit you didn't see coming. ADHD, stoicism, and the skill nobody talks about. ## [Claudedogged the blog](https://josh.miami/posts/claudedogged-the-blog) _2026-03-01 · josh.miami_ My blog was a piece of shit. I fixed it. Eight features, one session, three mistakes. I'm coining a term. ## [Recipe PDFs Ready Fast (Sponsored)](https://crawlproof.com/a/fI4j2oxTKF3Y) _2026-03-01 · **Sponsored**_ Turn online or personal recipes into clean, printable PDFs in seconds. ## [TokenGolf started with a simple question: what if AI cost had stakes?](https://josh.miami/posts/token-golf-origin) _2026-02-28 · josh.miami_ Claude Code already tracked cost. The weird part was that the number didn't really matter. TokenGolf started there. ## [Publishing blog posts from my phone with Claude and GitHub Actions](https://josh.miami/posts/publishing-blog-posts-from-my-phone-with-claude-and-github-actions) _2026-02-28 · josh.miami_ Three files, a GitHub label, and a Claude skill. The whole setup for going from post idea on my phone to live on the site without touching a code editor. ## [Two Claude Code tricks I actually use every day](https://josh.miami/posts/claude-code-profiles-and-audio-hooks) _2026-02-22 · josh.miami_ Profile switching for clean account separation, and audio notifications so you stop babysitting the terminal. Both take five minutes to set up. ## [The Model Is Not the Variable Anymore. The System Is.](https://josh.miami/posts/the-model-is-not-the-variable-the-system-is) _2026-01-22 · josh.miami_ The real shift was not becoming a power user. It was realizing the model was no longer the main variable. The surrounding system was. ## [I Canceled Cursor. Claude Code Is the Default.](https://josh.miami/posts/i-canceled-cursor-claude-code-is-the-default) _2025-10-09 · josh.miami_ Canceling Cursor was not a dramatic decision. The overlap phase was already over. Claude Code is the default and has been for a while. ## [Claude Code Was Winning. I Just Wasn't Listening.](https://josh.miami/posts/claude-code-was-winning-i-just-wasnt-listening) _2025-06-26 · josh.miami_ The switch was not clean. The old lane still existed, but the terminal was already taking the work that mattered most before I was ready to call it the default. ## [I Stopped Defaulting to OpenAI](https://josh.miami/posts/i-stopped-defaulting-to-openai) _2025-03-13 · josh.miami_ There is a real difference between sampling a model and repeatedly choosing it. This is the point where the preference stopped being theory and started becoming habit. ## [The Interface Was Solved. The Model Wasn't.](https://josh.miami/posts/the-interface-was-solved-the-model-wasnt) _2024-12-12 · josh.miami_ Once the surface stopped being the problem, the more interesting question became which model actually fit the work. That is where the preference started shifting.