aye-buddy: Claude Code sandboxer now keeps your sandboxed sessions separate from unsandboxed sessions.

CLAUDE.md should contain your core design choices you want agents to follow, not your project structure.

#TIL about #git export-subst

Sure, but I kind of fear there will be more of them.

There's already too many general purpose languages.

How many Claude tabs do you have in your browser? I just managed to hit 6.

I literally had to search through the tabs to see which one is making weird noises.

Spot the typo $ git branch contains c01450bb

how tf do I disable auto play on vidoes on snaps?

The Claude code was not really designed with the idea of running inside a sandbox.

Pointer is just a special case of a slice.

Something I haven't tried, because it was too time consuming.

Now that we have AI that can solve all the easy problems, we should move to solving harder problems instead of solving the same problems faster.

Trying to understand macvlan v bridge #networking and why the former disallow host->guest communication, is this good enough explanation?

I know, but I still don't like it.

Python? Yuck :), I always liked perl and ruby more, but maybe I'm just drawn to the weird languages.

Damn, Fable just used $60.72 + $56.66 in two subagent heavy sessions.

I love how you can do grep { defined } in #perl and it just works.

Is this idiomatic way of returning exit code from #python script or just AI slop? if name == "main": raise SystemExit(main())

it's intriguing and terrifying at the same time.