Hacker News: sebringj

New comment by sebringj in "I was interviewed by an AI bot for a job"

sebringj — Thu, 12 Mar 2026 00:00:20 +0000

so was I then I got the job... i think it's innovative because bots are just filling these out anyway, being mad at a tech company for building tech, that's like calling the kettle well a kettle.

New comment by sebringj in "Don't post generated/AI-edited comments. HN is for conversation between humans."

sebringj — Wed, 11 Mar 2026 23:12:52 +0000

I do too care about this but I say this in the reality in which we are. This reminds me of those signs "no shirt, no shoes, no service" except it's much worse, only sentient beings will actually care about it, while non-sentients will simply trample over the sign while token predicted laughter erupts from their token predicted sense of humor artifact.

Elon said it well, there must be some disincentive to do this.

New comment by sebringj in "The dead Internet is not a theory anymore"

sebringj — Wed, 11 Mar 2026 22:55:16 +0000

Isn't this more about identity crisis? Not in the psychological sense but in the internet sense? Who is real who isn't? Crypto proof of work idk? Your profile like d LinkedIn or Hacker News or something gives your clawbot x credits worth of legit automated queries or rate limits on your behalf? It could be flipped upside down where we don't spend our eyes off on reading websites anymore, it just goes and gets it for us but I may be hallucinating.

New comment by sebringj in "I made my favorite AI tool"

sebringj — Sun, 22 Feb 2026 19:21:34 +0000

ah right, thanks for commenting though, nice to see some life... once you submit they don't allow URL field to be adjusted.

https://sebringj.github.io/autonomo/

New comment by sebringj in "Absurdity of the Term "AI Engineer""

sebringj — Sun, 22 Feb 2026 17:59:22 +0000

i get it but what else do you call it right now? its like a full stack dev that builds chatbots with apis and streaming and context fixes, it's a confusing term because it hints at ML core but then you see what they want and there is no model training/inference usually involved. Maybe "AI Fullstack?" idk either but I literally have been doing this over the last 2 years but ya ended up doing model training/inference too but the IMO obtuse leadership didn't want it.

I made my favorite AI tool

sebringj — Sun, 22 Feb 2026 17:49:28 +0000

i do not submit things to hacker news unless its related to my favorite tool ever, literally, that i happened to have made. i made this out of being super lazy and wanted my copilot (works in all ai editors) to run my UI while its coding and validate it at the same time by using the apps. i don't know how to contain how good this is for me to use other than putting it here for people to look at. so using it with opus 4.5-4.6 its extremely good, however using it with gpt-5.3 its still good but you have to remind it to use the "autonomo help" when it forgets how to use it correctly sometimes.

anyways, please check it out if you are curious and want very fast efficient UI driven (multi app/web/desktop at the same time, agnostic) validation while you vibe. I just keep using it everyday but still waiting for something to just make this obsolete.

web page:

https://sebringj.github.io/autonomo/

github:

https://github.com/sebringj/autonomo

Comments URL: https://news.ycombinator.com/item?id=47113021

Points: 5

# Comments: 5

My AI now has a license to drive

sebringj — Mon, 16 Feb 2026 15:32:17 +0000

Built something small to stop the endless "it works on my machine" loop with AI assistants.

Autonomo lets your AI actually see the app state, drive UI elements across devices, and verify changes in real time—right from your copilot or cursor etc. No more blind code suggestions.

If you're tired of AIs that talk big but can't prove it on real hardware, give it a spin and let the AI take the wheel for a bit. This is for apps only though so it's optimized for speed, not generic vision or specific device.

https://sebringj.github.io/autonomo/ Curious if anyone else is experimenting with giving agents proper eyes/hands in dev loops.

Comments URL: https://news.ycombinator.com/item?id=47036273

Points: 2

# Comments: 0

New comment by sebringj in "I love validating E2E while vibing in my code editor, possible? Hell Yes"

sebringj — Sat, 14 Feb 2026 18:06:14 +0000

so I made this and frankly i was in shock that it worked so well, so i decided to make it plugable as an MCP, i have used this approach for 3 projects now and I frankly can't live without it. I hope other people find it useful but ya you never know. Coding in a vacuum isn't useful so lmk what you think if you try it out etc.

I love validating E2E while vibing in my code editor, possible? Hell Yes

sebringj — Sat, 14 Feb 2026 17:52:59 +0000

Article URL: https://sebringj.github.io/autonomo/

Comments URL: https://news.ycombinator.com/item?id=47016617

Points: 3

# Comments: 1

My AI now drives my apps and checks its own work in 50ms loops

sebringj — Fri, 13 Feb 2026 15:21:37 +0000

When you address your own problem and it works really well, that is when you go ok this actually could help other people.

My issue started as taking on a side project sports league app and web as it was insane to attempt as a solo dev – all-in-one sports community with social feed, league mgmt, live broadcasting, payouts for trainers/refs, travel/venue bookings, waze type live field conditions, AI assistant gluing it all together. web + ios + android + desktop as a solo dev. do everything to matter in this space in other words, every small tweak (chat sync, payment edge case, broadcast switch, booking conflict, field update prop, AI response consistency) meant i had to spin up browser + ios sim + android emu + desktop, play different roles (fan/player/parent/coach/ref/vendor/manager), manually poke everything, stare to see if sync held without desync/crash/security fuckups. There was just not enough time to get it done within a year if I had to manually check things worked every time.

I thought ok if AI coding can actually see and drive to check its work but in a meaningful deeper way, that would let it stop proclaiming "It worked!" when it didn't. What if it could actually know while its developing? In the past I worked with puppeteer, playwright for web and tried some mobile stuff that seemed very heterogeneous and even tried vision calls that was super expensive. None of those would be deep enough anyway, I needed one approach and those others were also slow or super expensive with vision, there had to be a better way. Why not JSON state that bubbles up server errors too with a local socket server making it fly?

I was in shock for 2 days that it worked so well. It just built out the apps in about a month as it could go build -> verify did it work? -> iterate (repeat) -> done... but this time done meant it actually tried it and it worked and could even interplay devices across web and mobile. So my problem is now trying to show my sports connected, non tech cofounder how to use it so he can give meaningful demos to prospective clients.

Meanwhile, it might just be worth your time to check out autonomo mcp.

https://sebringj.github.io/autonomo/ https://github.com/sebringj/autonomo

Hoping to see where it evolves with enough feedback as I tend to be a hermit in a vacuum.

Comments URL: https://news.ycombinator.com/item?id=47003691

Points: 1

# Comments: 0

New comment by sebringj in "Rari – Rust-powered React framework"

sebringj — Thu, 12 Feb 2026 21:38:11 +0000

ah gotcha... i was picturing like this massive web page that was dog slow that you just like made mad fast... but that's cool though. I'll run it.

I built autonomo on github, it's this mcp you plugin to your whatever ai editor and it drives the app/web/desktop whatever thing and validates it actually runs and works... keep building homie!

New comment by sebringj in "Rari – Rust-powered React framework"

sebringj — Thu, 12 Feb 2026 21:05:36 +0000

this looks neat and reminds me of "Ferrari" for fast and combines "react" so it's a great name. Was there a specific use case for this that inspired you? I would imagine some massive existing heavy thing that you plugged in to fix to save it? I just put my open source thing out so it's nice to see some traction on yours, rooting for you.

New comment by sebringj in "I started programming when I was 7. I'm 50 now and the thing I loved has changed"

sebringj — Tue, 10 Feb 2026 21:59:10 +0000

idk, i'm loving the newness of all of it, I feel more empowered than ever before, like it's my time. Before startups would take like a year to get going, now it's like a month or so. It's exciting and scary, we have no idea where it's going. Not boring at all. I was getting bored as shit and bam, now i can dream up shit quick and have it validated to, ya i figured that out with an MCP so ya this is my jam. Program MCPs and speed it up!!!!!!

New comment by sebringj in "Runtime validation is still fucked in AI coding agents"

sebringj — Tue, 10 Feb 2026 17:35:41 +0000

ya, for real, my boss was like let's do e2e testing with AI, look for solutions out there... then like 2 days later he's like wtf is this bill, and i was like you wanted that right? Was using vision calls in azure foundry and was like over 100 bucks or something just in 2 days of me setting it up and trying it out with all the test cases it had.

Runtime validation is still fucked in AI coding agents

sebringj — Tue, 10 Feb 2026 17:22:50 +0000

AI agents (Cursor, Claude computer-use, Copilot agent mode, etc.) have gotten stupidly good at spitting out code. Prompt → boom, clean code. The marketing says "it just works."

It fucking doesn't.

You run it in a real app and immediately hit the same bullshit wall every time: - Hallucinated logic only reveals itself under real data or edge cases - UI updates magically forget to sync across devices (mobile → web = sad trombone) - API calls quietly return 401s or other crap that gets swallowed in some lazy try-catch - Vision-based agents crawl like molasses (2–10s per action) and torch tokens like it's free - Background pings and unrelated fetches make it impossible to tell what actually caused what

I tried pretty much everything out there and none of it quite scratched the itch I had: fast, structured, cross-platform runtime visibility without vision bloat or having to wire up a ton of hooks.

Quick rundown of the usual suspects:

- Pure vision/computer-use (Claude 3.5/4, ADEPT-style): zero setup, works on anything — but latency from hell and token burn is brutal for anything longer than a demo - Playwright / browser MCP servers: fast and structured for web — but web-only, selectors shatter like glass, no native mobile - Appium + vision hybrids: cross-platform on paper — but still vision-dependent and setup is a pain - Sandboxed agents (OpenHands, SWE-agent): decent for repo tasks and shell stuff — not so much for live app UI/network state - Explicit hooks/bridges: precise when you bother adding them — but requires code changes, which sucks

Couldn't find anything that gave me low-latency structured JSON state (UI elements, network, errors, logs) across platforms, local-first, without the usual trade-offs. So yeah, I got fed up and built a small local MCP server to solve it for myself.

Full disclosure: it's called Autonomo MCP https://github.com/sebringj/autonomo — very early, just launched.

I don't usually do this "I built a thing" thing — my open-source contributions are mostly small fixes and PRs — but I honestly couldn't see a better way in the current landscape.

It is my hope that Anthropic (or someone) will eventually ship a clean native solution for this. They already fixed BM25 tool calling to shrink context like crazy; I'd love to see them (or the industry) make runtime validation "just work" out of the box too.

Sometimes when you code in a vacuum you think your shit smells good. lmk if I'm off base here, I grew up with a mean grandpa so I'm cool with it.

Comments URL: https://news.ycombinator.com/item?id=46963340

Points: 1

# Comments: 2

New comment by sebringj in "Show HN: Autonomo MCP – Developing while E2E Testing"

sebringj — Tue, 10 Feb 2026 15:12:33 +0000

Hey HN,

Tried posting this yesterday evening but timing was off and it got buried quickly (4 points, no comments). Reposting now during peak hours.

I built Autonomo MCP because AI coding agents (Cursor, Claude, etc.) are great at generating code but terrible at actually running and validating it in real apps. They hallucinate, rely on slow screenshots, or break on multi-device flows.

Autonomo MCP gives agents a fast (~50ms), structured JSON view of app state (semantic UI IDs, network calls, errors, logs) across web + iOS + Android + desktop simultaneously. Local-first, MCP-integrated, no cloud leaks.

- Swap vision bloat for tiny tokens - Cross-device validation (e.g. tap mobile → see web update) - Early stage, just launched: https://github.com/sebringj/autonomo

Would love feedback: Does this solve a pain point for anyone building agentic dev tools? What platforms/workflows would you want next? (Android/Kotlin and C#/.NET are on the TODO list.)

Thanks for any stars, comments, or tries!

– Jason

Show HN: Autonomo MCP – Developing while E2E Testing

sebringj — Tue, 10 Feb 2026 15:12:01 +0000

Article URL: https://github.com/sebringj/autonomo

Comments URL: https://news.ycombinator.com/item?id=46960739

Points: 2

# Comments: 2

Show HN: Autonomo – AI developing while E2E testing

sebringj — Mon, 09 Feb 2026 22:28:22 +0000

Hey HN,

- Swap vision bloat for tiny tokens - Cross-device validation (e.g. tap mobile → see web update) - Early stage, just launched: https://github.com/sebringj/autonomo

Would love feedback: Does this solve a pain point for anyone building agentic dev tools? What platforms/workflows would you want next? (Android/Kotlin and C#/.NET are on the TODO list.)

Thanks for any stars, comments, or tries!

– Jason

Comments URL: https://news.ycombinator.com/item?id=46952450

Points: 4

# Comments: 0

New comment by sebringj in "Tell HN: Announcing tomhow as a public moderator"

sebringj — Wed, 02 Apr 2025 19:11:28 +0000

Congrats Tom. I wonder if a particular model could be used as a baseline for these values or if they are already doing that to check first level prior to a human in the loop? I myself have been using AI for this purpose and have found it getting pretty good. I know its not a replacement for thoughtful moderation however, a tailored model for HN would also promote the tradition of HN in terms of having it not just be about who is there and have it more trained on its best practices to promote consistency, possibly as an aid.

New comment by sebringj in "GPT-4o"

sebringj — Mon, 13 May 2024 19:52:40 +0000

What struck me was the interruptions to the AI speaking which seemed commonplace by the team members in the demo. We will quickly get used to doing this to AIs and we will probably be talking to AIs a lot throughout the day as time progresses I would imagine. We will be trained by AIs to be rude and impatient I think.