Who am I? I'm a software engineer with over a decade of experience. I started on LinkedIn and now branching into other social medias. Here are a couple of free resources for software engineers👇

I can’t use Fable, it simply keeps changing me to opus 4.8 on almost everything

Still testing fable .. so far it’s been solid .. I think … idk my 5 hour limit ran out quick Will report more in 1h 24min

if fable is actually significantly better. than im ok with the double token usage.

Both are horrible metrics but tokenmaxing is easier to fake than LoC

Idk if it’s just me, but I really slept on Claude Cowork because of Claude Code. Cowork actually is really really good at nondev things that I had to originally build workflows into CC for

What’s a tool or library you cannot live without nowadays?

I am starting to think that using AI coding agents properly is more skill than luck

Spent the morning flipping worktree.baseRef. Default fresh branches sub-agents off origin so they never see my WIP. Set it to head and they branch off my current state, which is what I actually wanted on a 4-step plan where each step builds on the last.

The pattern I keep reusing for autonomous work: a daily cron that fires Claude Code with a fixed prompt, points it at one repo, and posts the output changes (last 24 hours) to my Slack channel. Keeps me updated from all that was merged the day before.

The more Claude Code and Codex create browser bots, the less I think users are going to be the ones interacting with productivity apps

Slack MCP is OP - I literally run everything through here

If you're designing a multi-agent system, the question I keep coming back to: should the coding agent be a destination app or a callable component? I keep going back and forth. I had Gemini as part of an automated workflow that is started from Claude Code and swapped it out to simply the API, but I'm thinking the agent may be the better option

So I keep hearing Codex 5.5 is the best out there, but most people are still on Claude Code (including myself) - is it really worth the switch?

No free coffee for you!

I just read an article that Anthropic planned for 10x growth in Q1. They got 80x annualized. AI is going to be the biggest thing that's ever happened.

A lot of Google hate lately, most of which is deserved. But I just wanna remind people there’s a VERY long list of people that took a Google internal tool or paper and later started an entire company out of it, by making it available to the public. If there’s problems at Google, they’re not with the people, they’re cultural. There’s a wide distribution of engineers obviously, so even dummies like me get lumped in with the genius level programmers, but I would never bet against Google.

Hot take, but I am not sure if "reviewing every single line of code" by Claude/Codex is needed if you have a strong CI/CD & testing pipeline in place

The Obsidian Claude Code setup I've been running for a couple weeks splits into three layers, and the split is what keeps the agent from clobbering source material. vault/ (root layer) 1. raw/ # sources I drop in, LLM read-only 2. wiki/ # compiled pages, LLM owns 3. CLAUDE.md # human-written rules Most builds I've seen skip the first two.

Can we take a moment to silently clap to both my children sleeping completely through the night!?

Pi is becoming crazy popular. 214 releases in 6 months, ~47k stars, single developer from what I know. The possibilities of what can be built in a small amount of time is wild.