Feed aggregator

Show HN: Claude Code for iPad – Agentic AI coding tool with file ops, Git, shell

Hacker News - Fri, 03/06/2026 - 3:00am

We built an agentic coding tool that runs on iPad. Claude reads your codebase, plans changes, edits files, and pushes to GitHub — autonomously, in a loop. 7 integrated tools (Read, Write, Edit, Glob, Grep, Bash, Git), all executing locally on the device.

The shell is a JS polyfill (~25 Unix commands with pipes, chaining, and redirection). Git uses isomorphic-git. API calls stream token-by-token via SSE.

We used it to develop itself — dogfooding from day one.

We've hit iPad platform limits: no persistent background processes, no real shell for builds/tests, iOS purges IndexedDB after 7 days. Looking for collaborators to solve the last-mile problem — especially anyone with experience in iOS hybrid apps, WebContainers, or keeping background servers alive on iOS.

Repo: https://github.com/M8seven/claude-mobile Detailed writeup: https://github.com/M8seven/claude-mobile/issues/1

Comments URL: https://news.ycombinator.com/item?id=47272260

Points: 1

# Comments: 0

Categories: Hacker News

Unlimited users, free and ad-free remote employee management tool

Hacker News - Fri, 03/06/2026 - 3:00am

NeuralOps

Comments URL: https://news.ycombinator.com/item?id=47272253

Points: 1

# Comments: 1

Categories: Hacker News

A claudeism that I want to confirm if anyone else is experiencing

Hacker News - Fri, 03/06/2026 - 2:57am

Does your Claude use the phrase "I contain multitudes" often? I've tried to search for it but the things I mostly get are LLM-centered people using it in their sentences online (duh, wonder where they learned that one from.....) and ironically a moltbook entry which further confirms it for me Makes you think that alongside the usual alignment and just throwing moar compute and redesigning the formulas underneath how much work is put into engineering it's personality. The stylometric way it replies and how for example Claude has the very claudeisms. Was this one simply learnt during the so divergent process of training or just manually wired into it as a random bit that makes it sound a little more sophisticated? I know that most people that'd ever have contact with this are under very strict NDAs but I'd be more than glad if someone could shed some light for my uninitiated I A shrink on payroll and some H100s is all it takes to go from soulless context oblivious Gemini to the very soulful Claude that can instantly mirror your tone but get his own when ready to argue? It's obviously a illusion sure but it's the very thing that I think separates Claude as more conscious or understanding for some people (......including me so I might be a bit biased....) A general discussion on AI personality would also be welcome

Comments URL: https://news.ycombinator.com/item?id=47272240

Points: 1

# Comments: 0

Categories: Hacker News

Show HN: Desktop Automation with Codex

Hacker News - Fri, 03/06/2026 - 2:56am

Tried using doing some desktop automation by sending codex screenshots and stepping through generated instructions. It's rough, but it (mostly) works. In the screenshot it accidentally presses 71336 haha.

Comments URL: https://news.ycombinator.com/item?id=47272234

Points: 1

# Comments: 0

Categories: Hacker News

Show HN: Making remote MCP servers handle local files and generated artifacts

Hacker News - Fri, 03/06/2026 - 2:51am

Most MCP examples look clean when the client and server run on the same machine.

That breaks once the MCP server is remote. Tools that expect local files cannot access them, and tools that generate files leave those artifacts stranded on the server.

I built remote-mcp-adapter to bridge that gap. It stages client-side files for upstream tools, captures output artifacts, and keeps normal MCP traffic flowing through unchanged.

Repo: https://github.com/aakashh242/remote-mcp-adapter

Would love feedback from people building remote or centralized MCP setups.

Comments URL: https://news.ycombinator.com/item?id=47272201

Points: 1

# Comments: 0

Categories: Hacker News

Show HN: Koshei AI – a voice-native AI language university (A1 to D2)

Hacker News - Fri, 03/06/2026 - 2:49am

Hi HN,

I’m building a voice-first AI language teacher in the browser using: - Next.js - Gemini Audio API for STT - Gemini TTS - Supabase

The product vision is a structured AI “language university” rather than a general chatbot.

Right now my biggest technical problem is: How would you implement lightweight browser-native lip sync for a static avatar image while TTS audio is playing?

I’ve tried: - Three.js + VRM (too heavy / unstable for this use case) - simple canvas mouth animation - CSS-only pulse effects

I want: - something realistic enough to feel alive - low dependency weight - web-compatible - stable on mobile

Secondary issues: - MediaRecorder reliability on mobile Safari - reducing transcript latency - voice UX for guided teaching rather than free-form chat

Demo: https://koshe-al.onrender.com

Repo: https://github.com/Bugsbuny24/Koshe-Al-

Would love technical suggestions, architecture criticism, or examples of similar systems done well.

Comments URL: https://news.ycombinator.com/item?id=47272188

Points: 1

# Comments: 0

Categories: Hacker News

Show HN: Free in-browser sprite sheet packer and slicer with JSON export

Hacker News - Fri, 03/06/2026 - 2:11am

Built this because I kept needing a quick sprite sheet tool while working on side projects and everything either required a download or had a paywall.

It runs fully in the browser — no upload to a server, no login. Supports: - Packer mode: upload separate frame PNGs, reorder them, export as atlas + JSON - Slicer mode: upload an existing sheet, define rows/columns, split into frames - JSON metadata compatible with Phaser, custom WebGL renderers etc.

Works with Unity and Godot workflows too. Would love feedback on what's missing.

Comments URL: https://news.ycombinator.com/item?id=47271940

Points: 1

# Comments: 0

Categories: Hacker News

Pages