🔊 Turn Codex Desktop Into an Intel Mac App

📦 miscend/codex-rebuilder
This repository is a small JavaScript tool that rebuilds the Codex Desktop app so it can run on older Intel-based Mac computers, even though the official app is packaged for Apple Silicon Macs. It solves a compatibility gap for people who still use Intel Macs and want the desktop app experience instead of only the command line. The script takes files from the official installer, downloads the matching Intel app framework, rebuilds required components, and creates a new `Codex_Intel.app`. That makes setup much easier and safer than hand-editing app files, and helps people keep using Codex on hardware that is no longer officially supported.
🔗 https://github.com/Miscend/codex-rebuilder
📰 https://news.ycombinator.com/item?id=47137484

🔊 Clash: Smart Traffic Router for the Open Internet

📦 clash-sh/clash
Clash is an open-source network traffic router and proxy tool that helps people control how apps connect to the internet. Instead of sending all traffic the same way, it can use rules to route different websites or services through different connections (such as direct access or proxy servers). This is useful for improving reliability, getting around regional network restrictions, testing connections, or managing privacy and speed tradeoffs in one place. It is popular because it is flexible, scriptable, and cross-platform, making advanced network control available through a single engine that many desktop and mobile tools can build on.
🔗 https://github.com/clash-sh/clash
📰 https://news.ycombinator.com/item?id=47137470

🔊 Axon Turns AI Coding Agents Into a Team

📦 axon-core/axon
Axon is an open-source tool that helps teams run AI coding assistants as automated workers instead of one-off chat sessions. You point it at a GitHub issue or task, and it can launch an agent in an isolated container, let it make code changes, and return a pull request. It solves the messy parts of managing many AI coding jobs at once, like setup, security boundaries, tracking progress, and chaining tasks together. This is useful for teams that want repeatable, scalable AI help for bug fixes, tests, docs, and maintenance work without manually supervising every step.
🔗 https://github.com/axon-core/axon
📰 https://news.ycombinator.com/item?id=47137491

🔊 AWSim: Test AWS Apps Without the Cloud

📦 sivchari/awsim
AWSim is a lightweight tool that imitates parts of Amazon Web Services (AWS) on your own computer. Instead of connecting to the real cloud for every test, developers can run common services like storage, databases, queues, and notifications locally. This solves a practical problem: cloud testing can be slower, cost money, and require accounts, network access, or setup that is hard to manage in team and automated test environments. AWSim makes development faster and safer by letting people build and test cloud-based apps offline, repeat results more easily, and catch problems before deploying to real AWS.
🔗 https://github.com/sivchari/awsim
📰 https://news.ycombinator.com/item?id=47137716

🔊 Add Crypto Features Fast with Starkzap

📦 keep-starknet-strange/starkzap
Starkzap is an open-source TypeScript toolkit that helps app developers add Bitcoin-related and other crypto features to web, mobile, and backend apps through Starknet. Instead of stitching together separate wallet, token, and transaction tools, developers get one SDK for onboarding users, checking balances, sending tokens, staking, and even supporting gasless payments. It aims to hide much of the usual blockchain setup friction, such as complicated wallet flows and fee handling, so apps can offer a smoother experience to everyday users. This is useful for teams that want to launch crypto-enabled features quickly without building low-level blockchain plumbing from scratch.
🔗 https://github.com/keep-starknet-strange/starkzap
📰 https://news.ycombinator.com/item?id=47137736

🔊 Your AI Agent, Now With a Rewind Button

📦 flytohub/flyto-core
Flyto Core is an open-source engine that helps people build and run multi-step AI automations more reliably. It gives an AI assistant hundreds of ready-made tools for common jobs like browser actions, file handling, data parsing, APIs, and scheduling, then runs those tasks in a structured workflow. The big problem it solves is visibility: instead of ending with only a chat log, you get step-by-step records, state snapshots, and the ability to replay a workflow from a specific step when something fails. That makes AI-driven tasks easier to debug, repeat, audit, and turn into dependable automations for real work.
🔗 https://github.com/flytohub/flyto-core
📰 https://news.ycombinator.com/item?id=47137952

🔊 One Search Bar for Your Whole Workplace

📦 getomnico/omni
Omni is an open-source workplace search and AI assistant platform that helps teams find information spread across tools like Google Workspace, Slack, Confluence, and Jira in one place. Instead of hunting through separate apps, employees can search once, ask questions, and get answers based on the documents and messages they already have permission to access. It is especially useful for organizations that care about privacy, because it can run in their own cloud and keeps existing app permissions in place. Omni also supports different AI providers, so teams can choose how to power the assistant without being locked into a single vendor.
🔗 https://github.com/getomnico/omni
📰 https://news.ycombinator.com/item?id=47138046

🔊 Run Notion From Your Terminal

📦 4ier/notion-cli
notion-cli is a command-line app that lets you use your Notion workspace from a terminal instead of the web app. It covers everyday tasks such as browsing pages and databases, creating and updating content, managing blocks, and handling comments or users through simple commands. It also supports shell auto-completion and an AI agent mode for automation. The problem it solves is speed and repeatability: clicking around in a browser can be slow, especially for repeated tasks. This tool is useful if you want to automate Notion work, connect it to scripts, or manage content in a faster, keyboard-first workflow.
🔗 https://github.com/4ier/notion-cli
📰 https://news.ycombinator.com/item?id=47133849

🔊 Opal: Build-Your-Own AI Coding Assistant

📦 matteing/opal
Opal is an open-source AI coding assistant built in Elixir that works in two ways: as a command-line tool you can use on your projects and as a library you can plug into your own apps. It helps with practical coding work like reading and editing files, running shell commands and tests, and debugging issues, which reduces repetitive tasks and constant context switching. That makes it useful for developers who want to speed up everyday work, automate parts of software development, or experiment with creating their own coding assistants. It also supports parallel helper agents and live visibility into what the agent is doing.
🔗 https://github.com/matteing/opal
📰 https://news.ycombinator.com/item?id=47138518

🔊 Docker Setup, Minus the Headache

📦 unslop-xyz/noodles
Noodles is an open-source command-line tool that helps developers spin up a ready-to-use project environment with Docker, without hand-writing lots of setup files. It looks at your codebase (for example, Python, Node, Rust, or Go), picks a sensible base image, and generates the Docker Compose setup for you. This solves a common problem: getting a new app running locally often takes time, breaks on different machines, and requires container know-how. Noodles makes that setup more automatic and consistent, so teams can onboard faster, avoid environment mismatches, and spend more time building features instead of plumbing.
🔗 https://github.com/unslop-xyz/noodles
📰 https://news.ycombinator.com/item?id=47138656

🔊 Turn Command Lines Into Real Apps

📦 charmbracelet/bubbletea
Bubble Tea is an open-source toolkit for Go developers to build interactive command-line apps that feel more like full interfaces than plain text scripts. Instead of hand-coding screen updates, keyboard input, and app state from scratch, it gives a clear pattern for handling user actions and drawing the screen, with support for mouse input, colors, and fast performance. This solves a common problem: terminal apps get messy and hard to maintain as they grow. It is useful because teams can build polished tools, dashboards, and workflows inside the terminal faster, with more consistent behavior and less low-level terminal work.
🔗 https://github.com/charmbracelet/bubbletea
📰 https://news.ycombinator.com/item?id=47138688

🔊 Browse Open Data Without the API Headache

📦 noi-techpark/opendatahub-databrowser
This repository contains the Open Data Hub Data Browser, an open-source web app that helps people explore public datasets, especially tourism and regional data from South Tyrol, in one place. Instead of digging through many API endpoints and technical documentation, users can browse clear categories such as events, accommodations, weather, webcams, and locations, then inspect records and jump to related API or Swagger pages. It solves the common problem of knowing what data exists and where to find it by turning a complex data platform into a searchable, easy-to-navigate interface. That makes it useful for developers, researchers, and curious non-experts before building apps, dashboards, or reports.
🔗 https://github.com/noi-techpark/opendatahub-databrowser
📰 https://news.ycombinator.com/item?id=47138797

🔊 FPScanner: DIY Bot Detection for Websites

📦 antoinevastel/fpscanner
FPScanner is an open-source JavaScript library that helps websites detect bots and suspicious automated traffic by collecting a short-lived browser fingerprint and checking for common automation clues. Instead of forcing teams to buy an expensive, all-in-one anti-bot service, it provides self-hosted building blocks they can connect to their own server. It can spot signals from popular browser automation tools, compare results across different browser contexts, and send protected payloads with timestamps and random values to make replay or fake submissions harder. This makes it useful for defending signup, login, and scraping-targeted pages while keeping more control over data and detection rules.
🔗 https://github.com/antoinevastel/fpscanner
📰 https://news.ycombinator.com/item?id=47139031

🔊 Turn Your Mac Into a Local AI Hub

📦 jundot/omlx
oMLX is a macOS app and local server that lets people run AI language models directly on Apple Silicon Macs, with a menu bar control and browser dashboard instead of command-line setup. It solves a common headache: local AI tools can be hard to manage, slow to reload, and awkward when switching between models. oMLX adds easier model downloads, automatic model loading/unloading, and caching that saves work to SSD so repeated conversations can resume faster, even after restarts. It also speaks OpenAI- and Anthropic-style APIs, so existing apps and coding tools can connect to it with minimal changes, making private, on-device AI much more practical.
🔗 https://github.com/jundot/omlx
📰 https://news.ycombinator.com/item?id=47139158

🔊 Turn Messy SQL Into Clickable Maps

📦 buva7687/sql-crack
SQL Crack is an open-source Visual Studio Code extension that turns database query text (SQL) into interactive diagrams. Instead of reading long, hard-to-follow queries line by line, users can see how data moves through filters, joins, calculations, and results, then click around to trace where columns come from. It also scans a project folder to map relationships between files and tables, highlight possible performance issues, and show what might break if something changes. This helps analysts, engineers, and learners understand complex data logic faster, debug mistakes sooner, and explain query behavior visually with exports and side-by-side comparisons.
🔗 https://github.com/buva7687/sql-crack
📰 https://news.ycombinator.com/item?id=47139578

🔊 A Tiny AI Assistant You Can Actually Read

📦 nicktikhonov/bitclaw
BitClaw is a small, open-source personal AI assistant that lets you chat with Claude through Telegram while keeping the assistant isolated in a Docker container on your computer. It is designed for people who want AI help with notes, coding, and scheduled tasks, but do not want to trust a huge, hard-to-audit codebase. Because the project is only about 1,500 lines of TypeScript, you can realistically read it, understand how it works, and customize it for tools like Gmail or Calendar. That makes BitClaw useful as a safer, more transparent starting point for running your own AI assistant.
🔗 https://github.com/NickTikhonov/bitclaw
📰 https://news.ycombinator.com/item?id=47136566

🔊 Caddy: the web server that makes HTTPS easy

📦 caddyserver/caddy
Caddy is an open-source web server and app platform that helps people put websites and web apps online with less setup pain. Its standout feature is automatic HTTPS, which means it can get and renew security certificates for you instead of making you handle that manually. It also supports reverse proxying, load balancing, and serving static sites, so it can sit in front of apps and route traffic reliably. For developers, small teams, and hobby projects, Caddy is useful because it simplifies secure hosting, reduces configuration mistakes, and works well across local machines, cloud servers, and containers.
🔗 https://github.com/caddyserver/caddy
📰 https://news.ycombinator.com/item?id=47139913

🔊 Plug-In Skills for AI Coding Helpers

📦 huggingface/skills
Hugging Face Skills is a collection of ready-made instruction packs that teach coding assistants how to handle common Hugging Face tasks, such as creating datasets, training models, running evaluations, and publishing research outputs. Each skill bundles step-by-step guidance, helper scripts, and templates so an AI agent can do a job more reliably instead of guessing. The repository also works across multiple agent tools, including Codex, Claude Code, Gemini CLI, and Cursor, which reduces setup friction and lock-in. It is useful because teams can start from proven workflows, automate repetitive machine-learning work, and customize or contribute new skills instead of building everything from scratch.
🔗 https://github.com/huggingface/skills
📰 https://news.ycombinator.com/item?id=47139902

🔊 Find Your Lost Glasses With Your Phone

📦 yjeanrenaud/yj_nearbyglasses
This repository contains an Android app called NearbyGlasses, built to help people locate Bluetooth-enabled smart glasses (especially XREAL Air glasses) when they are nearby. Instead of guessing where you left them, the app scans for the glasses and uses signal strength to give a rough sense of distance, making it easier to track them down around your home, office, or travel bag. It also shows basic battery information and connection details. The project is useful because it solves a simple but frustrating everyday problem: misplaced wearable devices that are hard to spot, especially when they are small, folded, or buried in a case.
🔗 https://github.com/yjeanrenaud/yj_nearbyglasses
📰 https://news.ycombinator.com/item?id=47140042

🔊 LoMux: FFmpeg Power Without the Command Line

📦 zblauser/lomux
LoMux is a lightweight, cross-platform desktop app that makes video and audio conversion easier for everyday users. Instead of typing long FFmpeg commands, you use a simple interface to convert files, run batches, edit media details, and watch progress in real time. It can also download and convert YouTube videos in one step when used with yt-dlp. The project solves a common problem: powerful media tools are often confusing and text-heavy. LoMux puts those features behind clear presets and buttons, making it useful for creators, students, and anyone who needs quick format changes without learning command-line syntax or installing a bulky media program.
🔗 https://github.com/zblauser/LoMux
📰 https://news.ycombinator.com/item?id=47140207

🔊 One App to Run an AI Coding Team

📦 generalaction/emdash
Emdash is an open-source desktop app that helps teams run several AI coding assistants on the same software project at once. Instead of trying one tool at a time, it gives each agent its own safe workspace, lets them work in parallel, and then shows the code differences side by side so a person can compare and merge the best result. It also connects tasks from GitHub, Linear, and Jira, so work can move from issue to finished code faster. This solves the bottleneck of managing multiple AI agents manually and makes AI-assisted development easier to organize, review, and trust.
🔗 https://github.com/generalaction/emdash
📰 https://news.ycombinator.com/item?id=47140322