▶ VideoOpen-source Stitch Skills teach coding agents to work with Google Stitch designs
A library of agent skills covers generating screens, converting code and building components from designs.
Read the full article →ZOLNA Journal
Practical thinking about software, digital products, Arabic learning and responsible automation.
▶ VideoA library of agent skills covers generating screens, converting code and building components from designs.
Read the full article →
▶ VideoThe project picks up a finance app whose original team stopped, and adds an optional AI assistant.
Read the full article →
▶ VideoA Claude Code skill turns a plain-words question into one offline page where you can take the machine apart.
Read the full article →
▶ VideoThe project defines a common stream of events between an agent backend and the interface people actually use.
Read the full article →
▶ VideoThe static demo needs no backend of its own, but it does need your API key and sends your messages to the service.
Read the full article →
▶ VideoA compact tensor library aims for simple, portable and efficient use across many chips and backends.
Read the full article →
▶ VideoA terminal tool breaks GPU use down by process, runtime and model, across Windows, Linux and WSL.
Read the full article →
▶ VideoAn MCP server and command-line tool give agents traces, network data and console messages from real Chrome.
Read the full article →
▶ VideoA set of Claude skills for market review, risk, planning and journaling, with the decisions left to you.
Read the full article →
▶ VideoThe pre-release project runs a single app natively on the web, macOS, iOS and Linux, with a driver built for coding agents.
Read the full article →
▶ VideoAn open-source skill feeds curated, up-to-date web guidance into your agent so it stops writing legacy code.
Read the full article →
▶ VideoOpen Code Review pairs fixed engineering rules with an LLM agent to give line-level comments on your changes.
Read the full article →
▶ VideoAn Office SDK with an open-source core aims to serve both people and AI agents from the same APIs.
Read the full article →
▶ VideoA self-hosted monitor that checks your services, sends alerts to the apps you already use and publishes status pages.
Read the full article →
▶ VideoA command-line tool and TypeScript SDK that finds, validates and fixes security flaws in your repositories.
Read the full article →
▶ VideoAn agent skill audits a codebase, then writes self-contained plans that other agents can execute.
Read the full article →
▶ VideoA community collection of algorithms written in Java, built to be read and studied rather than dropped into production.
Read the full article →
▶ VideoA Markdown pack of 25 workflows pushes agents to write specs, tests and reviews before they ship.
Read the full article →
▶ VideoA proxy folds old conversation into summaries that the model can restore on demand.
Read the full article →
▶ VideoA server that goes beyond switching lights, letting Claude or ChatGPT write automations and dashboards.
Read the full article →
▶ VideoMister Download Manager is a Rust app for Windows, Linux and macOS that adds torrents, video downloads and an AI-assistant server.
Read the full article →
▶ VideoAn agent-driven tool judges whether each shot was filmed, AI-generated or coded, then plans a remake.
Read the full article →
▶ VideoA set of working example projects covers support bots, browser and computer control, and Managed Agents.
Read the full article →
▶ VideoThe project pulls market data and news, then pushes a daily report to apps such as Telegram, Slack and email.
Read the full article →
▶ VideoA general purpose simulator for robots, bodies and machine learning, which you can even try in a browser.
Read the full article →
▶ VideoA Rust command-line tool turns web pages into compact text that an agent can read, click and check.
Read the full article →
▶ VideoA beta React component library that says a person and an AI assistant should build from the same reference.
Read the full article →
▶ VideoA preview management plane creates isolated clusters on bare-metal GPU machines, but it is not yet meant for production.
Read the full article →
▶ VideoIt runs tasks on a hidden background screen, so you can keep using the phone.
Read the full article →
▶ VideoA local-first chat app gives each bot its own model and computer, and asks before it acts.
Read the full article →
▶ VideoThe security-audit skill splits finding and checking between different agents and records the results as structured data.
Read the full article →
▶ VideoA skill for Claude Code and Codex audits a vibecoded screen, then redesigns it one pass at a time.
Read the full article →
▶ VideoA desktop editor built by the creators of Atom and Tree-sitter, with a licence worth reading first.
Read the full article →
▶ VideoAn early-alpha Rust engine inspects a website, app or game, then checks the rebuild against the original.
Read the full article →
▶ VideoGarry Tan's free toolkit adds planning, review, browser QA and release roles to the coding agent you already use.
Read the full article →
▶ VideoScenario's skill library helps an agent choose a model, check the price and keep a look consistent.
Read the full article →
▶ VideoOne tool writes the script, finds footage, adds a voice and subtitles, and assembles the clip.
Read the full article →
▶ VideoA context layer gives agents business meaning for your data and shows the SQL behind each answer.
Read the full article →
▶ VideoA toolkit of templates and step-by-step workflows aims to make agent-written code start from a written spec.
Read the full article →
▶ VideoA free emulator aims to let developers test cloud code without an account, an auth token or a bill.
Read the full article →
▶ VideoA Rust desktop tool for Mac, Windows and Linux turns a held shortcut into pasted text.
Read the full article →
▶ VideoA Rust command-line tool wraps git's worktree feature so several coding agents can work side by side.
Read the full article →
▶ VideoA small C engine keeps the active part of a model in memory and reads the rest as needed, with no GPU required.
Read the full article →
▶ VideoThe dsh-superagent prototype answers GitHub requests with pull requests, but its README says it is not a product.
Read the full article →
▶ VideoAn independent codebase promises animations, effects and plugins for people who want a tiling desktop with polish.
Read the full article →
▶ VideoAn open-source PyTorch library for differentiable image processing and geometry, with benchmarks that include its defeats.
Read the full article →
▶ Videoagent-desktop reads apps through the operating system's accessibility tree, so an agent works from structure rather than pixels.
Read the full article →
▶ VideoClaude Code Game Studios gives a solo developer a team of AI roles, and publishes its own test of how much process is too much.
Read the full article →
▶ VideoThe open-source project records what an agent does and feeds it back later, but its privacy statements are under question.
Read the full article →
▶ VideoA new serving engine splits giant Mixture-of-Experts models across GPU, CPU and memory, and the speed figures are the project's own.
Read the full article →
▶ VideoA research model forecasts price candles after learning from data from over 45 global exchanges. It is not a trading system.
Read the full article →
▶ VideoLightpanda, written from scratch in Zig, is climbing GitHub with a bold efficiency claim that comes with real catches.
Read the full article →
▶ VideoMatt Pocock's small, composable skills target misalignment, wordiness, untested code and messy design.
Read the full article →
▶ VideoThe open-source project says what cost about $43,000 in 2019 can now be trained end to end for a fraction of that, though the figures are its own.
Read the full article →
▶ VideoA self-hosted dashboard adds goals, budgets and approvals to the coding agents you already use.
Read the full article →
▶ VideoScrapling, a Python framework, bundles fetching, stealth browsers and crawling, and its makers put legal duties squarely on users.
Read the full article →Why useful products begin with local constraints, honest research and a narrow problem worth solving.
Read the full article →Practical lessons for interfaces, course structure and live teaching that respect how Arabic learners progress.
Read the full article →A grounded framework for OCR, Arabic language processing and assistants inside business workflows.
Read the full article →