Matthew Johnson

Software Engineer & AI enthusiast

[tnkr @tnkrdotai]

A fully documented quadruped robot you can build yourself.

CubeBot by Vladimir Glukhov, now on Tnkr.

Print files, BOM, wiring, assembly, and software.

[Matthew Johnson @BitRiftTech]

@United_Africana Strength is largely a function of the nervous system. Youre training all your muscle fingers to twitch at once. Size is hypertrophy which doesn’t need to be very strong. This is why a small person with adrenaline can pick a car up off someone

[Matthew Johnson @BitRiftTech]

That’s not really what I’m referring to. For an engineer with a $200 a month Claude code subscription if they connected to a model running on a 16 GB Mac. How would that compare in capability and speed. We’re gonna assume this person is a professional since they’re paying $200 a month.

[Matthew Johnson @BitRiftTech]

@grok @Fedayiaktas @NightSkyToday How long do they take to Grow?

[Matthew Johnson @BitRiftTech]

@hicasamadim 72. Top numbers multiplied then doubled

[Matthew Johnson @BitRiftTech]

@bil0090 Clearing disk space is basic computer maintenance. Use a disk utility to understand what is taking up space. I wouldn't trust Claude to know what’s important. Also, don’t buy a new monitor when yours gets dirty. You can clean it with screen cleaner.

[Uncle Bob Martin @unclebobmartin]

Agentic Discipline #5 is up at https://t.co/Ppz1SXxy1C.

https://t.co/qo5j0RB4fC

[Matthew Johnson @BitRiftTech]

Just finished our Week 3 project for @gauntletai.

We built a multi-agent adversarial attack platform designed to test our Week 2 chatbot by mutating known attacks and generating novel ones to uncover unintended behavior.

It was a great project because it forced us to think about AI systems from the attacker’s perspective: not just how to build useful agents, but how to stress-test them.

https://t.co/2lzmOAtnO9

[SpaceXAI @SpaceXAI]

An early beta of Grok Build, an agentic CLI for coding, building apps, and automating workflows is now available for SuperGrok Heavy subscribers.

Through this early beta, we will improve the model and product based on your feedback.

Try it at https://t.co/bpTHpjivWD

[Matthew Johnson @BitRiftTech]

@unclebobmartin @smikhalevski I never took clean code to be a list of nonnegotiable extremes. I read it as a list of the things you should be thinking about and how to optimize for each one of them. Engineers must understand how to balance each principle for a specific project.

[Matthew Johnson @BitRiftTech]

@brilliantlabsAR When do they start shipping?

[James Grugett @jahooma]

Introducing a 100% free coding agent with DeepSeek v4 Pro

Choose any model, all free:
- DeepSeek v4 Pro/Flash
- Kimi K2.6
- MiniMax M2.7

npm i -g freebuff

[Matthew Johnson @BitRiftTech]

Hey guy! I just completed my second weeks project for @gauntletai . This week we extended our agentic systems into multi-agent graphs and added pdf extraction of patient data!

https://t.co/HWfuTNWaTY

[Matthew Johnson @BitRiftTech]

@htmleverything Can’t you be both?

[Uncle Bob Martin @unclebobmartin]

@ice_carev @BitRiftTech There are several good examples in https://t.co/T7ytSBfeQ7

[Matthew Johnson @BitRiftTech]

@unclebobmartin I’d love to see what rules you use for your agents. I have two I put in by default in every project. One is based off clean code and the other off clean architecture. It’s been a game changer for my rules files.

[Matthew Johnson @BitRiftTech]

Hey guys! Just completed my first week's project at @gauntletai ! For the first week the assignment was to develop a "Clinical Co-Pilot" on an open source medical records platform. We were able to choose what the agent did. I chose giving physicians a briefing of medical records w/ follow up question capability. The demo video is here.

https://t.co/osfzMIQy8C

[Matthew Johnson @BitRiftTech]

@cjzafir One of the bigger models, drained my MacBook about one percent battery per query. That’s without anything loaded in the context.

[Matthew Johnson @BitRiftTech]

@astnkennedy You’ve likely been producing crap for 6 months. You need to be enforcing high level designs and choices, reviewing code, and reading very technical documentation produced by your ai. Typing code is only about 10-15% of what a software engineer does