Mauricio Scheffer

Just another developer.

[rob dezendorf @robdezendorf]

1/ STOP paying for CLAUDE CODE and API CREDITS

use an ai code editor powered by Chipotle's customer support bot instead.

Introducing Chipotlai Max — OpenCode but with a hardcoded burrito-brain. Does Pepper 1 write sloppy code? yes. but it's flavorful.

[Mauricio Scheffer @mausch]

@championswimmer I have currently multiple basic tier subscriptions:

- Opencode Go
- Github Copilot Pro
- Claude standard team seat
- ChatGPT plus

best bang for the buck atm is ChatGPT I think (used to be Copilot)

[Mauricio Scheffer @mausch]

Congrats to Atlassian on getting jira to load faster than github in 2026.

Sadly, not because jira got faster.

[Shriram Krishnamurthi (primary: Bluesky) @ShriramKMurthi]

Startled to find out that there are young people who haven't read James Iry's magnificent "A Brief, Incomplete, and Mostly Wrong History of Programming Languages". Please drop what you're doing and head over.
https://t.co/bnzifwQ2p9

[Mauricio Scheffer @mausch]

@aramh Unsolicited recommendation: Factorio

[Armin Ronacher ⇌ @mitsuhiko]

More musings after some people got upset about the word clanker. https://t.co/gXPC6iRP0g

[Mauricio Scheffer @mausch]

Cloned some repo to try to contribute something and just had to delete their .claude dir, it had the most obnoxious skills

[Mauricio Scheffer @mausch]

Hell is other people's prompts and skills

[Mauricio Scheffer @mausch]

Jfc am I the only one still working on two or three $100 1080 monitors?

[Gergely Orosz @GergelyOrosz]

My current work monitor has issues, so looking for a new one. I could get a replacement Dell 27-inch one for ~$600, or an Apple Studio XDR 27-inch for $3,000.

I was thinking: would this 5x as expensive monitor help me do better work *at all*?

I see parallels with AI spend...

[Mauricio Scheffer @mausch]

@thdxr Very unfortunate.
Personally I label something as slop when it's missing some obvious context about the problem to be solved or topic being discussed, which the human driving the AI either didn't articulate at all or didn't even bother to check in the output before publishing.

[Mauricio Scheffer @mausch]

Same with gmail
A few days ago I just wanted to try out something with the gmail MCP server...
Look at all this nonsense to set it up... https://t.co/Ir1PNLDn4F
Did that only to be greeted with an "unauthorised" on the first operation. Who knows where the hell things went wrong.

[Mauricio Scheffer @mausch]

This so much.
Partner struggled to set up a google ads campaign. Unnecessarily complicated UI compared to e.g. instagram.
I think there's got to be some specific AI feature for this already.
There is, but apparently it's just a read-only chatbot 🤦‍♂️

[bone @boneGPT]

You need to stop benchmaxxing and start productmaxxing. Why compete in their arena of gameable benchmarks at all?

Nobody cares how it did a percent better on AgentBenchBotMaxArcAgi3. Play in the arena you win.

Example: Why isn't AI Gmail solved yet??? I sometimes get an autofilled post. Sometimes I don't. None of it is clear.

Why can't I add memories and rules to my email? Why aren't there AI templates I can choose from for every reply that are based on what I would send? Who the hell is in charge over there?

This should be the FIRST THING. I should have 3 response templates autogenerated for every email to pick from. I should have rules and memories. It should be obvious.

Focus.

[Mauricio Scheffer @mausch]

So I think fuck it, I'll just get Opencode to drive this with API calls.
Which sounds great on paper until you get sucked into the hopeless black hole of configuring permissions on GCP...

[Mauricio Scheffer @mausch]

Pin all your dependencies (easier now, not always so trivial 10yr ago)
Don't update dependencies for the sake of it.
You must have a reason for updating.
Update intentionally.
If you do update you must go through all changelogs.

I must have said this 1000 times in my career.

[Mitchell Hashimoto @mitchellh]

Fork your dependencies, trim them to only your use case, never update unless it breaks for your users. I’ve been vocal about this for 10+ years. I’ve always said that updating is way riskier than latent bugs (which can be tracked and CVEs monitored).

If you are updating a dependency, it’s on you to analyze every single commit in the full transitive set of dependencies. If you dont see anything compelling, dont update!

I remember at HashiCorp once in awhile an engineer would try to update a dep or replace a DIY lib with an external one and id always ask “show me the commit we need.” Dont update for the sake of it.

Feeling pretty swell about this mentality with all the supply chain attacks happening.

[Mauricio Scheffer @mausch]

@jacobCole4000 @kunchenguid Lots of people fed up with Anthropic for quite some time now.
Vote with your feet i.e. cancel your Claude subscription and use OpenAI or Opencode Go or whatever.
It's the beauty of having a healthy market!

[Mauricio Scheffer @mausch]

@maheshb Yep things are moving quickly but there's still so much to do!

[Mauricio Scheffer @mausch]

@maheshb LLMs are tools... we drive them to generate the code we want, and similarly we drive them to catch those things they can't do on their own, or use them as helpers e.g. ask questions about the code to speed up your own understanding

[Mauricio Scheffer @mausch]

@Kyriakos_Pelek yes, problem is when the context in brownfield is often scattered around multiple knowledge bases, or across multiple repos, or even not written anywhere, and the changes often need to be surgical. Ralph works best as a slop cannon IME

[Mauricio Scheffer @mausch]

For my personal greenfield projects I find myself converging on this workflow:

- ralph for bootstrapping
- tweak environment, tooling
- tweak some features
- ralph for some big feature
- tweak, tweak, tweak
- another ralph
- etc

[Mauricio Scheffer @mausch]

For brownfield I still find it difficult to fit ralph... too much implicit context (my fault for not making it explicit tbh), divergence in a loop means slop on steroids :)