Ephes Blog

Why This Matters: Coding Agents Are Changing How I Work

Coding agents like Claude Code are quite powerful, but watching them work can be like watching paint dry—especially when you're just clicking "approve" repeatedly. Running them with --dangerously-skip-permissions speeds things up, but doing that on your main development machine? That's asking for trouble.

My solution: dedicate an old Mac mini as a sacrificial lamb for agent experiments. But this created new challenges that took me down a rabbit hole of SSH, tmux, and shell configuration.

Gardening, not architecture --Oblique Strategies

It's been a hectic week both at work and in my personal life - haven't had time for much else. I'm thinking maybe I should write a blog post about wrapping Django web apps in Electron shells 😀. Also, as of last week, Claude Code is now available with Pro/Max subscriptions, not just through the paid API. Since Claude Code has become my go-to tool for coding with LLMs, I'm considering upgrading to Max.

Articles

• claude-trace | might be useful to find out how this thing works..
• DjangoconEU2025: Talk recommendations
• HTML5 Elements You Didn't Know You Need | Need to remind myself from time to time those elements exist..
• Preserving referential integrity with JSON fields and Django | Interesting approach. I'd probably try something like this if I had to implement galleries again. In django-cast, what I do is sync the JSONField with the many-to-many relations on every save. I'm also not sure it's a good idea to make the database schema dependent on the application though.
• How Anthropic teams use Claude Code | Okay, this is mostly marketing fluff disguised as a case study, but I do like seeing how different people approach automating their workflows with Claude Code.
• The last six months in LLMs, illustrated by pelicans on bicycles
• Open Source as a business owner