Ma.dev

NimoteCode is for developer.

[Ma.dev @mobiledevlab]

AI coding from your phone — not just a remote terminal.

NimoteCode lets you run a real remote coding workflow:
Inspect → Plan → Execute → Verify → Preview.
Your dev machine does the heavy lifting. Your phone stays in control.

#AICoding #MobileDevelopment #CodingAgent #RemoteDevelopment #NimoteCode

[Ma.dev @mobiledevlab]

Got a bug report today from a NimoteCode user about SSH private key authentication.

Importing a key from a file can fail on Android, and manually entered keys may not persist correctly after reconnecting.

Thanks for the report — I’m working on a fix for the next release. 🙏

Honestly, this made me even more excited than getting my first subscriber.

It means someone is actually using NimoteCode seriously enough to hit real problems and take the time to report them.

That feels like progress.
#NimoteCode

[Ma.dev @mobiledevlab]

I used to think the coding model was the main factor.

Then I started building my own coding agent.

The same model can feel completely different depending on the agent around it.

Tool calling, context management, retrieval, task state, loop detection, error recovery, verification, and even the stopping conditions all change the result.

Two agents can use the exact same model, but one gives the model precise code context and verifies every change, while the other gives it noisy context and lets it keep guessing.

After dozens of steps, their trajectories can be completely different.

That changed how I think about AI coding agents:

The model sets the capability ceiling.
The runtime determines how much of that capability you can actually realize.
A coding agent isn't just an LLM + tools.

It's an entire system designed to make the model converge on the right result.

I wrote more about what I learned building one 👇

[Ma.dev @mobiledevlab]

"Why the Same AI Model Can Feel Completely Different Across Coding Agents" by @nimotecode #DEVCommunity https://t.co/fbgLlmvMG1

[Ma.dev @mobiledevlab]

I’ve been comparing mobile dev tools lately, and they’re actually solving very different problems.

Termius → SSH / SFTP / server management
Blink Shell → terminal-first + Mosh
Redock → mobile SSH for AI coding agents
Orca → desktop agent environment + mobile companion
ServerCC → persistent SSH + Git Diff + Tailscale
Moshi → agent-aware SSH / Mosh terminal
NimoteCode → editor + files + terminal + Git + AI + SSH in one mobile workspace

So I don’t think the real question is:
“Which one is best?”

It’s:

**What do you actually want to do from your phone?**

For me, the interesting split is becoming:
**terminal-first vs agent-control vs full mobile IDE.**

Curious what people here actually use 👇
What’s your current mobile coding setup?

#MobileDevelopment #AICoding #DevTools

[Ma.dev @mobiledevlab]

@MSiekmoeller that's still impressing

[Ma.dev @mobiledevlab]

"Why the Same AI Model Can Feel Completely Different Across Coding Agents" by @nimotecode #DEVCommunity https://t.co/fbgLlmvMG1

[Ma.dev @mobiledevlab]

@Surendar__05 X is the best one. reddit is too strict for marketing

[Ma.dev @mobiledevlab]

"get the best market data"—that might be the hardest part....

[AGTrader @agtrader]

Listen to me very carefully....

Get the best mkt data... pair it with Astra or the next frontier model, and you have about a year to make FU money before all of this turns into a giant AI gamed shitshow.

[AGTrader @agtrader]

Listen to me very carefully....

Get the best mkt data... pair it with Astra or the next frontier model, and you have about a year to make FU money before all of this turns into a giant AI gamed shitshow.

[Ma.dev @mobiledevlab]

First rule: don’t pick Mark Zuckerberg.

He’ll probably just fire you and build it himself. 😂

[kapilansh @kapilansh_twt]

you’re starting a startup

choose your co-founder

[Ma.dev @mobiledevlab]

AGI isn’t here yet.

What we have today are incredibly powerful systems built largely around next-token prediction, search, and reinforcement learning.

Scale them up enough, and they can look like they’re thinking.

And honestly, they’re already better than most humans at a lot of things — coding, planning, verification, and more.

But being extremely capable isn’t the same as having a real world model, continuously learning from experience, or reasoning the way humans do.

The output can look remarkably human.

The process underneath is fundamentally different.

These models are insanely useful. But useful ≠ AGI.

[Greg Brockman @gdb]

an important essay on the state of AI and the choices and challenges ahead: for openai, the field, and the world.

we're in the AGI era, and need to approach the next few years with seriousness, thoughtful deliberation, and navigating the collective challenges together. we can make the future so much better than the past.

[Ma.dev @mobiledevlab]

While AI has vastly improved efficiency, it has also intensified competition and made the landscape harder to control. Although AI enables the rapid completion of tasks, these capabilities are accessible to everyone; when everyone’s capabilities are boosted equally, it effectively means that no one has gained a relative advantage.

In the past, the tech industry might have been dominated by those with superior intellectual prowess, but now that AI has raised the intellectual baseline for the majority, the nature of competition is shifting from the mental to the physical realm.

I believe this is precisely why things are becoming increasingly difficult.

#AI #TechJobs #Developer #Competition #FutureOfWork #SoftwareDevelopment #CareerAdvice #AIImpact

[fidexCode @fidexcode]

It was easy to make money in tech.

But now, you will suffer before you make a dime.😭

[Ma.dev @mobiledevlab]

@rain8miao ok, I'll take ur advice and use astra medium 😀

[Ma.dev @mobiledevlab]

@theo When the market is flooded with software wrritten by AI, can software development still be profitable for solo developer？I was depressed for that.