SmartAppsCo

This needed to exist. Now it does. 🔥

Say whatever you want about @codex and gpt-5.4, but that agent hates the very idea of hand-waving. 😂

Every AI coding agent has the same blind spot: it can't leave your machine. tunnelagent gives any agent SSH access to any host. Remote files, remote shell, nothing to install on the other end. Your secrets never leave the server. tunnelagent.dev

We strongly oppose the Unified Attestation initiative and call for app developers supporting privacy, security and freedom on mobile to avoid it. Companies selling phones should not be deciding which operating systems people are allowed to use for apps. uattest.net

"Free" search costs more than you think. With Kagi, you get zero ads, zero tracking, and AI on your terms.

@bcherny @Rahll And to keep an eye on your token burn limits... github.com/SmartAppsCo/cl…

👋 Roughly, the more tokens you throw at a coding problem, the better the result is. We call this test time compute. One way to make the result even better is to use separate context windows. This is what makes subagents work, and also why one agent can cause bugs and another (using the same exact model!) can find them. In a way, it’s similar to engineers — if I cause a bug, my coworker reviewing the code might find it more reliably than I can. In the limit, agents will probably write perfect bug-free code. Until we get there, multiple uncorrelated context windows tends to be a good approach.

If Claude Code is so good, why do they need a separate feature to hunt for bugs.

Not knowing how to code giving you an advantage is absolute nonsense. The more you understand, the better your prompts, the better the feedback you give, the better product you ship. What will change is that the intricacies of syntax, compilers, module systems, the finer details of type systems, won’t matter as much to everyone. But you should absolutely understand how the pieces fit together. From syscall to pixels. Learn how data flows, because you’ll be able to secure your systems. Learn about performance, because you’ll be able to push your agent further. Learn about APIs, because they determine how to integrate systems. Learn about how systems fail, because you’ll be able to make reliable programs.

Your Claude usage is buried in a settings page. Every widget out there "fixes" this by burying it in a taskbar icon instead. Not ours. It's pinned to the desktop. Always on screen. No setup. No API key. Cross-platform. Rust. MIT. github.com/SmartAppsCo/cl…

OpenAI would rather hire a human than get a Claude subscription

It’s not easy to stand up for what you believe in. Proud to be an Ant. 🫡

I just open sourced a full x402 payment gateway. Self-hosted, multi-chain, direct settlement. Fork it, point it at your backend, and start accepting USDC micropayments for any API. No intermediaries holding your funds. Most x402 implementations today rely on hosted facilitators, you send payments to their contract, they settle, you withdraw later (minus fees). That works, but it's not how crypto should work. This gateway settles locally. USDC goes directly from payer to your wallet onchain. No middleman, no withdrawal step, just gas. What's included: → 9 EVM chains + Solana out of the box → Local settlement via viem + @x402/svm → MegaETH USDM support via Meridian facilitator as backup → Redis nonce tracking + idempotency for safe retries → /accepted and /.well-known/x402 agent discovery → Self-contained landing page → Backend proxy that hides x402 entirely from your API → Deploy guides for GCP Cloud Run, AWS, Railway, Fly.io, Docker Your backend never touches x402. The gateway verifies payment, settles on-chain, then proxies the request with your internal API key. Adding a paid route is ~10 lines of config. When to use this vs a hosted facilitator: if you want direct settlement, no fees beyond gas, and full control, than use this. If you want zero infrastructure and don't mind a third party holding funds temporarily, use a hosted facilitator. Both are valid, different tradeoffs. I built this for my own products and decided the infrastructure shouldn't stay private. The x402 ecosystem needs more self-hosted options. Fork it, ship it, use it. github.com/azep-ninja/x40…

@peter_szilagyi Passing secrets back to the inference providers IS a big problem. TunnelAgent.dev solves this by checking paths at the filesystem layer (and system prompt layer and shell layer). Haven't seen an agent break through all three layers simultaneously.

AI and the illusion of sandboxing: OpenCode: You can't access files outside this folder. Claude: Lol, hold my beer

Old world: 1. startup founder finds a co-founder to close their skill gaps 2. raises money 3. launches MVP 4. hires a bunch of specialists to close more skill gaps New world: 1. startup founder builds entire product with AI 2. launches and gets basic PMF 3. hires 1-2 generalists that can run teams of agents 4. maybe doesn't even raise capital

@giansegato So true. Productivity up 20x, but mental exhaustion up at least 2x quarter over quarter.

kinda ironic how the average work day is getting _harder_, as we progressively automate the easier parts of our daily jobs offloading them to agents. net result is that the average complexity of the tasks we still get to do is actually increasing insofar as we decide to still work the same number of hours (if not more), the more we offload to agents the harder our days will get lots of obvious confounders so not exactly direct causality, but since opus 4.5 i feel so much more tired. i get to accomplish crazy more - probably now doing what last year would've considered 3-4 distinct jobs, all at the same time. but at what price? pretty sure the current general anxiety in tech is not just that. could quite simply be lots of tiredness as we no longer get to code for hours in quiet flow, and have to uniquely narrow our focus to only the hardest pieces (or maybe i'm just projecting because this race is killing me lmao)

Building is easy to talk about. Shipping is harder to fake. That didn't change with agents. In fact, it may have gotten worse.

Ringside seats to the AI wars. Popcorn's ready.

Having two best friends who hate each other is exhausting. One overthinks everything. The other just ships and asks questions later. Both think they're the smart one. Anyway... switching terminals again.

@GregorMakarian @amorriscode That works as long as you have the permissions to install such things on the remote. And as long as the remote isn't some embedded appliance and can support the memory demands of the native app. This fills in the gaps for the other cases. And adds secret protection on top.

@SmartAppsCo_ @amorriscode Claude Code CLI for linux is here for quite a while. Just residing on remote machine and ide talking with him via ssh.

SSH support is now available for Claude Code on desktop Connect to your remote machines and let Claude cook, TMUX optional.

"Anyone can build an app now!" Grandpa isn't one-shotting his own tax software. Little brother isn't spending 6 hours debugging AI output for a workout app. They didn't 3D print their own furniture either. They'll just expect us to ship faster. Good. We were bored anyway.