Just a collection of thoughts with a bit of sarcasm about side projects, web development and leadership written by a human. Ex-Nomad (37+ Countries, 18% π)
jsifalda.bsky.social
Joined February 2009
- Tweets4.5K
- Following1K
- Followers579
- Likes2.8K
Pinned Tweet
The #1 reason your team resists change has nothing to do with the change itself.
It's because learning something new makes them feel stupid.
kind of sad how seamless it was to submit an mcp to claude code store, all done by claude code itself.
meanwhile chatgpt plugin submission is so buggy i ended up talking to support.
says i need to be a verified dev, but i am.
still haven't managed to submit it.
#buildinpublic
there is something really weird about hr using ai for slack responses.
feeling like you're talking to a chatbot in situations that matter just doesnt feel right.
dont do it as a manager, and definitelly not as hr
my definition of a good codebase changed recently:
it's easy to change without breaking things.
that means guardrails everywhereβ¦
so agents (and humans) stay in line, and you can trust them to make changes without you policing it.
bit closer to standardization and unification π
We're donating Agent Skills to the Agentic AI Foundation (@AgenticAIFdn), the same foundation we donated MCP to.
Since December 2025, 70+ agent platforms have integrated the standard, including ChatGPT, Cursor, GitHub Copilot, and VS Code.
Here's what that means for you:
i kinda love my claude project called "running beast"
comments like this are gold:
"the hard thing for you is not distance. it is running slow enough to feel stupid. do that."
why dev productivity is hard to measure:
we judge it day by day, but the real impact of good or bad work shows up months or years later.
those decisions are hard to spot, and even harder to reward
no dashboard will tell you this
had to introduce a "no ai harness day" for myself as an eng. manager.
a day where i dont run any agents (at least until late). it was eating up my strategic thinking time.
its just easier to talk to an agent rather than think deeply.
some good ideas on inner vs outer loops and building trust in agents.
how to eventually remove the biggest bottleneck in development: yourself.
youtu.be/MN9dGgmLyso
btw if you dont want to listen the whole episode...
claude.ai/artifact/E76UnevZfβ¦
sending custom html reports with visualizations of the issues we discussed/encouraged has become the new standard in our team.
no need to build looker dashboards when claude code is much faster. ofc you need data somewhere, but a /production db is usually the best option
its weird how easy it is to build custom telegram bots mixed with n8n workflows using claude code (via claude -p on a vps).
even tho its against their tos, right?
(not using it for real, just a learning experiment...)
George Sifalda ππ§ retweeted
Get your Apple Reminders in Claude Code.
RemCTL 2.1.0 now ships with a Claude Code plugin and mod, so you can always see what's left to do today, or in a specific list. There's even a sidebar!
Get the update here: github.com/viticci/remctl
@amorriscode
I got a leadership lesson from my dad today.
I realized thereβs something I donβt do enough: showing people that theyβre doing a good job and have something to be proud of, even when things arenβt perfect.
Thatβs how you build the motivation to keep getting better as a team.
thx, d.
Simple trick for βfixingβ ownership:
forbid the team to say "they". Make them use "we".
"They didn't run the DB backup" -> "We didn't run the DB backup"
"They didn't test before release" -> "We didn't test before release"
Moves the problem from an outside group back to the team.
