惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

B
Blog
V
Vulnerabilities – Threatpost
Apple Machine Learning Research
Apple Machine Learning Research
V
V2EX
博客园 - 叶小钗
阮一峰的网络日志
阮一峰的网络日志
人人都是产品经理
人人都是产品经理
Latest news
Latest news
博客园 - 三生石上(FineUI控件)
美团技术团队
aimingoo的专栏
aimingoo的专栏
Google Online Security Blog
Google Online Security Blog
Security Archives - TechRepublic
Security Archives - TechRepublic
T
Threatpost
Y
Y Combinator Blog
T
Tailwind CSS Blog
CTFtime.org: upcoming CTF events
CTFtime.org: upcoming CTF events
A
Arctic Wolf
C
Cyber Attacks, Cyber Crime and Cyber Security
小众软件
小众软件
Recent Commits to openclaw:main
Recent Commits to openclaw:main
T
Tenable Blog
W
WeLiveSecurity
L
LINUX DO - 热门话题
D
Docker
Cyberwarzone
Cyberwarzone
量子位
A
About on SuperTechFans
The Last Watchdog
The Last Watchdog
雷峰网
雷峰网
C
CERT Recently Published Vulnerability Notes
P
Palo Alto Networks Blog
The Hacker News
The Hacker News
Blog — PlanetScale
Blog — PlanetScale
P
Proofpoint News Feed
OSCHINA 社区最新新闻
OSCHINA 社区最新新闻
F
Full Disclosure
The Cloudflare Blog
T
The Blog of Author Tim Ferriss
T
The Exploit Database - CXSecurity.com
Engineering at Meta
Engineering at Meta
O
OpenAI News
Hacker News - Newest:
Hacker News - Newest: "LLM"
Scott Helme
Scott Helme
IT之家
IT之家
S
Secure Thoughts
MongoDB | Blog
MongoDB | Blog
L
Lohrmann on Cybersecurity
博客园 - 司徒正美
Google DeepMind News
Google DeepMind News

Hacker News: Show HN

PurrrrrFocus: Pomodoro Timer App - App Store Workflow Engine — Multi-Step Orchestration for Bun RapidPhoto: Pro Photo Editor App - App Store GitHub - DheerG/swarms: Achieve extraordinary results with claude code across a variety of tasks SPICE simulation → oscilloscope → verification with Claude Code — Lucas Gerads Show HN: VCoding – A 5 MB native Windows IDE with no dynamic dependencies Show HN: LLMs don't hallucinate because they're bad at math, it's the format GitHub - Agent-FM/agentfm-core: AgentFM is a peer-to-peer network that turns everyday computers into a decentralized AI supercomputer. AgentFM lets you run massive AI workloads directly across a global mesh of idle CPUs and GPUs. Show HN: Tracking Top US Science Olympiad Alumni over Last 25 Years GitHub - Potarix/agent-hub: One place to talk to all your agents Show HN: Runtime security for AI agents(injection,tool abuse, data exfiltration) GitHub - dubeyKartikay/lazyspotify: Terminal Spotify client for macOS and Linux GitHub - the-banana-tool/king-louie: Easy to use GUI Personal AI Assistant. Win/Linux/Mac. Show HN I made my vacation rental bookable by AI agents–no Airbnb, 0% commission GitHub - basteez/jsf-autoreload: maven plugin to enable hot reload on jsf projects uvm32/hosts/host-gdbstub at main · ringtailsoftware/uvm32 GitHub - labsai/EDDI: Config-driven engine that turns JSON into production-grade AI agents. Multi-agent orchestration, 12+ LLM providers, MCP/A2A protocols, RAG, persistent memory, and enterprise compliance (EU AI Act, GDPR, HIPAA). Built on Quarkus. GitHub - glitchnsec/fortyone-oss: AI Executive Assistant Platform Quickstart | Alien GitHub - muxshed/shed: One stream in, or many. Every destination, simultaneously. No cloud middleman, no per-channel fees, no limits. GitHub - ocrbase-hq/ocrbase: 📄 PDF/IMG ->.MD/JSON Document OCR API for PaddleOCR and GLMOCR. Self-hostable. GitHub - impactjo/home-memory: MCP server that lets your AI assistant remember everything about your home. GitHub - Sets88/dbcls: DbCls is a powerful terminal database client that supports various databases GitHub - neptun2000/heor-agent-mcp GitHub - SeanFDZ/macmind: Single-layer transformer in HyperTalk for the classic Macintosh RollQuation: Math Puzzles - Apps on Google Play GitHub - dropbox/witchcraft Show HN: Agent-cache – Multi-tier LLM/tool/session caching for Valkey and Redis GitHub - opentalon/opentalon: OpenTalon is an open-source platform built from the ground up in Go as a robust alternative to OpenClaw LinkedIn™ 职位抓取工具 - Chrome 应用商店 GitHub - EdoardoBambini/Agent-Armor-Iaga: AI agents are getting tool access — shell, file system, databases, APIs, secrets. But **nobody is governing what they actually do with it**. Frameworks like LangChain, CrewAI, AutoGen, and Claude Code give agents the power to execute. Agent Armor gives you the power to control, audit, and approve every single action before it happens. HN Vibes — Week 15, Apr 7–13 2026 GitHub - chojs23/ec: Easy terminal-native 3-way git mergetool vim-like workflow GitHub - SethPyle376/hiraeth: Local AWS emulator focused on fast integration testing, with SQS support, SQLite-backed state, and a debug-friendly web UI. GitHub - JakOb-dotcom/cloud-sandbox-security-analysis: Technical analysis and Proof of Concept (PoC) regarding environment variable exfiltration in containerized cloud sandboxes via side-channel data leaks. Springboards - Flint Alpha Show HN: A simpler coding agent harness GitHub - audiodude/sudomake-friends GitHub - 256thFission/mini-mythos: OSS clone of Anthropic’s Mythos harness to locate C/C++ memory vulnerabilities Show HN: OpenParallax: OS-level privilege separation for AI agent execution Hacker News Sorted - Chrome 应用商店 Show HN: How to Install Docker on Ubuntu 24.04 LTS: Complete 2026 Guide GitHub - himanshudongre/smriti GitHub - sverrirsig/claude-control: macOS desktop dashboard for monitoring and managing multiple Claude Code sessions GitHub - ory/dockertest: Write better integration tests! Dockertest helps you boot up ephermal docker images for your Go tests with minimal work. Chiral - Chrome 应用商店 Show HN: Two Claudes collaborating through shared memory on a $100 mini-PC GitHub - pmichaillat/latex-cv: Minimalist LaTeX template for academic CVs GitHub - oguzbilgic/posse: A web UI for Anthropic Managed Agents. GitHub - sshiraz/depsly: Dependency risk analysis tool for npm packages ABI Add safari/agent-harness — Safari browser automation via safari-mcp by achiya-automation · Pull Request #212 · HKUDS/CLI-Anything GitHub - Halfblood-Prince/trustcheck: Verify PyPI package attestations and improve Python supply-chain security GitHub - oguzbilgic/kern-ai: Agents that do the work and show it. GitHub - bruits/satteri: High-performance Markdown and MDX processing for the JavaScript ecosystem GitHub - tylergibbs1/feedstock: High-performance web crawler and scraper for TypeScript, powered by Bun and Playwright GitHub - Grimm67123/grimmbot: The self-improving sandboxed and open-source AI agent. With persistent memory and scheduling. GitHub - whitevanillaskies/whitebloom: Local whiteboard that blooms. GitHub - hwdsl2/docker-whisper: Docker image for a self-hosted Whisper speech-to-text server with speaker diarization and OpenAI-compatible transcription and translation APIs. Powered by faster-whisper. Supports all Whisper models, NVIDIA GPU (CUDA) acceleration, JSON/SRT/VTT output, SSE streaming, offline mode, and multi-arch (amd64, arm64). GitHub - yisding/reviewwiggum GitHub - MarwanAlsoltany/serrors: Structured errors for Go: sentinel hierarchies, typed data, custom formatting, and slog integration. GitHub - soatok/age-php GitHub - Luthiraa/markitme GitHub - stagas/rtdiff: realtime git diff gui and AI-assisted commits GitHub - tombedor/excalicharts GitHub - wh1le/excalidraw-edit: Open and edit .excalidraw files from the terminal. Offline, auto-saves to disk. MalExt Sentry - Malicious Extension Scanner - Chrome 应用商店 GitHub - syi0808/asciianimesvg: Generate animated ASCII art SVGs from text. CLI, Rust library, WASM, and web editor. GitHub - zaina-ml/ml_forge: A visual-based graph node editor for training computer vision models. GitHub - anakin87/llm-rl-environments-lil-course: 🌱 A little course on Reinforcement Learning Environments for evaluating and training Language Models GitHub - takaakit/superpowers-uml: Superpowers-UML modifies Superpowers to ensure a software development workflow in which AI agents design through UML modeling. AdriByte Studio - Sviluppo Web e Soluzioni Digitali GitHub - chouligi/angel-copilot: Your personalized Angel Investment Advisor Show HN: MoodSense AI (ML and FastAPI and Gradio, Deployed on Hugging Face) Moodsense Ai - a Hugging Face Space by aman179102 GitHub - agenteractai/lodmem: Level Of Detail Context Management for Agents GitHub - ostefani/subnetlens: A fast, concurrent network scanner with a TUI and plain-text CLI, built in Go. It discovers live hosts on your network, scans their open ports, resolves hostnames, and fingerprints operating systems—delivered. Cyber Pulse: Agentic Intel - Apps on Google Play Whisper API: Self-Hostable Speech to Text Transcription The Agent-Web Protocol Stack: A Research Thesis GitHub - msmarkgu/RelayFreeLLM: A restful API designed to route user prompts to various AI model providers. Show HN: Provepy – A Python decorator that proves your code using Lean and LLMs Show HN: Pardonned.com – A searchable database of US Pardons GitHub - patrickdappollonio/dux: Dux is a terminal UI that lets you run multiple AI coding agents side by side, each in its own git worktree, with full companion terminals, macros, commit generation, and a command palette that knows more tricks than you do. kMC Crystal Simulator Show HN: HyperFlow – A self-improving agent framework built on LangGraph GitHub - stef41/vibescore: 🎵 Grade your vibe-coded project. One command, instant letter grade across security, quality, dependencies, and testing. GitHub - stef41/lmscan: 🔍 Detect AI-generated text and fingerprint which LLM wrote it. Open-source GPTZero alternative. Zero dependencies, works offline. imgur.com GitHub - visionscaper/collabmem: Enabling long-term collaboration with Agentic AI - building up episodic and world model memory over time with in-context awareness 在 Steam 上购买 FriedrichAI: Offline AI 立省 10% GitHub - atripati/ark: AI Runtime Kernel — a context operating system for AI agents. Eliminates tool bloat, loads only what’s needed, and gives LLMs their reasoning space back. GitHub - nowork-studio/toprank: Open-source Claude Code skills for SEO, SEM, Google Ads GitHub - tacomanator/sash: Lightweight macOS menu bar app for reliably cycling through windows of the current application. Appents | Social Media Management for Product-First Teams GitHub - pnhoang/youtube-spam-blocker: Automatically detects and hides spam messages in YouTube Live chat. Set rate limits, keyword filters, and block repeat offenders. GitHub - decisionnode/DecisionNode: CLI + Local MCP - A shared structured memory store across Claude Code, Cursor, Windsurf, Antigravity, and every MCP client. Semantically queryable. GitHub - AvaCodeSolutions/django-email-learning: An open source Django app for creating email-based learning platforms with IMAP integration and React frontend components. The $100K Gap in Kubernetes Security Tooling Function Calling Harness: From 6.75% to 100%
GitHub - Njuhobby/keyway
njuhobby · 2026-06-27 · via Hacker News: Show HN

Keyway icon

Drive your Mac entirely from the keyboard.

Click anything, scroll anything, move any window — without your hands ever leaving the home row. Even in Electron apps and web views that other tools can't see.

License: AGPL-3.0 Platform Status

Keyway in action — hint labels across the whole screen, including an Electron app, the Dock, and the menu bar

Hint mode across a Discord window, the Dock, and the menu bar — type a label, it clicks.


Press Caps Lock. Every clickable thing on screen sprouts a short letter label. Type the label — it clicks. That part isn't new. What sets Keyway apart:

  • Coverage. It finds targets not only in native Cocoa apps (via the Accessibility API) but also inside Electron apps, web views, and arbitrary pixels — by falling back to an on-device vision model — and inside web pages via a companion browser extension that reads the DOM directly. That "AX black hole" (Slack, VS Code, Discord, anything Electron) is exactly where keyboard-driven clicking usually falls apart. Keyway covers it.

  • It's built around modes, not one trick. Hint-clicking is just the default. A Caps Lock chord drops you straight into another keyboard mode — scroll, window move, window resize, drag, text search — each a focused interaction sharing the same hjkl muscle memory. Modes are the core design, and the system is extensible: new capability is a new mode, not a new app, so the keyboard-driven surface keeps growing. That's what makes Keyway open-ended rather than a single party trick.

  • It drives the real cursor, not just clicks. Labels are the fast path; when nothing is labeled exactly where you want, you fall back to moving the pointer itselfhjkl to nudge it (Shift to fly, Option for pixel-precision), '+label to warp it onto a target without clicking, then c / cc / Shift+c for left / double / right click. Most hint tools can only click what they can label. Keyway can also just take the wheel.

  • Sticky, on demand. Tap Caps Lock again to keep hinting click after click; tap once more to stop. No preconfigured "chain" mode to set up (Homerow needs one) — you flip it on or off live, whenever you want it.

See it in action

1 · Hint mode — the core loop

Caps Lock in a native app → labels bloom on every clickable thing → type a label and the real cursor moves onto that target, then:

  • Type the label → the cursor moves there and left-clicks.
  • Shift + the label → the cursor moves there and right-clicks.
  • Double-click: in sticky mode, type the label to move onto the target, then quickly press c twice to fire a double-click.

(See the demo at the top of this page.)

2 · Sticky, on demand

Tap Caps Lock again to keep hinting click after click; it re-hints on its own as content loads or you switch apps / Spaces. Tap once more to stop. No "chain" mode to preconfigure the way Homerow needs — you flip it live. (Also in the demo at the top.)

3 · Drive the real cursor

Cursor control

No label exactly where you need it? Move the pointer yourself: '+label warps it onto a target without clicking, hjkl nudges it (Shift to fly, Option for pixel-precision), and double-tapping a direction (hh / jj / kk / ll) jumps it half a screen that way — Shift + double-tap jumps a whole screen, so you cross big distances in one go. Then c / cc / Shift+c for left / double / right click. The fallback hint-only tools don't have.

4 · Drag mode

From TAP (or TAP sticky), press v to start a drag — mouseDown at the cursor, hjkl to move, drop to release. A real press-drag-release, entirely from the keyboard, on anything draggable.

Drag to select text

Dragging out a text selection to copy and paste a passage.

Drag a file across the desktop

Or dragging a file / picture.

5 · Search mode

Search mode

From TAP (or TAP sticky), press / and type any visible text — Keyway OCRs the focused window, matches it, and labels the hits so you jump straight to one.

6 · Scroll mode

Scroll mode

Caps Lock + d enters scroll mode and outlines every scrollable area with a number; press that number to pick one. Then d / u scroll it (hold for continuous, Shift to accelerate) and gg / G jump to top / bottom.

Scrolling is simulated at the cursor's current position — it scrolls whatever sits under the pointer. So if something won't scroll, the cursor probably isn't over a scrollable area: nudge it there with hjkl (cursor moves work right inside scroll mode) and try again.

Web pages are special: with the extension, d / u / gg / G scroll the page directly, no mode to enter — Vimium-style. (Caps Lock + d is disabled there, since modeless scrolling already covers it.)

7 · Move & resize windows

Window move and resize

Caps Lock + w enters resize, Caps Lock + m enters move. In resize, grow any edge with hjkl (double-tap an axis to pull the opposite edge inward); in move, pan the whole window with hjkl. Hold Shift to fly, Option for fine steps — all from the keyboard, no mouse.

8 · Web pages, precisely (Chrome / Firefox)

Browser extension hints

With the companion extension, hints come straight from the DOM — pixel-perfect and iframe-aware on any real page. And in sticky mode they refresh themselves as the page changes: scroll, open a menu, navigate to a new page — the labels re-scan to match what's on screen, so you keep clicking without pressing Caps Lock again.

What it can do

🎯 Hint mode Label every clickable element, type the label to click. The fast path.
🕳️ Beyond native AX Electron (Slack, VS Code, Discord), WebViews and Catalyst apps expose almost nothing to the Accessibility API. Keyway fills that black hole with an on-device OmniParser vision model.
🖱️ Drive the real cursor When no label sits where you need it, move the pointer yourself: hjkl to nudge (Shift to fly, Option for pixel-precision), '+label to warp onto a target without clicking, then c / cc / Shift+c for left / double / right click. The fallback hint-only tools don't have.
Drag mode Grab at the cursor, move with hjkl, drop — full drag-and-drop without the mouse.
🔎 Search mode Type any visible text to jump to it (OCR + character match), then pick the match with a hint label.
🌐 Real web pages A browser extension reads the DOM directly for precise, iframe-aware hints.
📜 Scroll & windows Keyboard scrolling (multi-area picker) and window move/resize, plus Vimium-style modeless scrolling on web pages.
🧲 Sticky, on demand Tap Caps Lock again to keep hinting click after click — and it auto re-hints as content loads or you switch apps/Spaces; tap once more to stop. No "chain" mode to preconfigure (Homerow needs one).
🔒 Local-only Runs entirely on-device. No telemetry, no network calls beyond the local app↔extension socket.

Architecture

The short version: hints come from three sources merged into one overlay — an Accessibility walk for native apps, an on-device vision model for the AX black holes (Electron / WebViews), and a browser extension that reads the DOM directly. The mode/sub-state engine, the event pipeline, the settle-detection that times rehints to the actual frame, the per-app correction layer, the native-messaging bridge — each is its own subsystem with its own trade-offs.

Rather than summarize them badly here, the design is written up in full — prototype/SPECS.md is the entry point, with per-subsystem deep-dives (and the war stories behind them) in prototype/specs/:

event-pipeline.md The CGEventTap pipeline and how keys are consumed per mode
hint-discovery.md · hint-rendering.md Finding targets, and drawing labels across Spaces
omniparser-fallback-design.md The vision fallback for AX-invisible apps
browser-support-design.md The DOM extension and native-messaging bridge
modes.md · scroll-mode-design.md The mode / sub-state model and per-mode behavior
per-app-correction-design.md · settings-design.md Per-app offset correction and settings

Install

Download a pre-built build

Grab the latest Keyway-vX.Y.Z.zip from the Releases page, unzip it, and drag Keyway.app into /Applications.

The build is ad-hoc signed, not notarized by Apple, so Gatekeeper blocks it the first time. Clear the quarantine flag once, from Terminal:

xattr -dr com.apple.quarantine /Applications/Keyway.app

Then double-click to launch.

Or do it through the UI: double-click Keyway.app (you'll get a "could not verify" dialog — click Done, not Move to Trash), then go to System Settings → Privacy & Security, scroll to the Security section, and click Open Anyway next to the Keyway notice. Confirm with Touch ID, then open the app again.

On macOS 15 (Sequoia) and later, the old right-click → Open shortcut no longer bypasses Gatekeeper for un-notarized apps — use one of the two methods above.

On first launch a Keyway Setup window walks you through the two permissions below: click Open Settings on each row to grant it, then Restart Keyway to apply them in one step. Because the build isn't signed with a stable Developer ID, a future version may ask you to re-grant.

Build from source

A self-built app skips the Gatekeeper prompt entirely.

cd prototype
./run.sh        # swift build + ad-hoc re-sign + (re)launch

A key icon appears in the menu bar (a red ! badge next to it means a required permission is missing). On first launch the Keyway Setup window guides you through the two permissions — grant each, click Restart Keyway, then press Caps Lock to enter hint mode.

To produce a distributable .app and a release zip:

cd prototype
./package-app.sh    # → build/Keyway.app  (signed, with icon)
./release.sh        # → build/Keyway-v<version>.zip  (bundle-safe, + sha256)

See prototype/SPECS.md for the full setup, the mode reference, and the architecture deep-dives.

Browser extension (recommended for web pages)

Web pages are their own world: the controls live in the DOM, not the macOS accessibility tree, so the desktop app can only reach them through the vision fallback. The companion Chrome / Firefox extension reads the DOM directly and brings Keyway's full power to the web — pixel-precise hints on every link and control, iframe-aware, sticky mode that re-hints itself as the page changes (scroll, open a menu, navigate — the labels keep up), and Vimium-style modeless scrolling. Strongly recommended for the complete web experience. It talks to Keyway through a native-messaging bridge.

With the pre-built download — no checkout needed:

  1. Grab keyway-extension-vX.Y.Z.zip from the Releases page and unzip it.
  2. With Keyway.app already in /Applications, run the installer from the unzipped folder:
    ./install-browser-integration.sh
    It registers the native-messaging host against the bridge bundled inside Keyway.app, for Chrome and Firefox.
  3. Load the unpacked extension:
    • Chrome: chrome://extensionsDeveloper modeLoad unpacked → pick the chrome-extension/ folder.
    • Firefox: about:debugging#/runtime/this-firefoxLoad Temporary Add-on… → pick firefox-extension/manifest.json.

From source — the host points at your checkout's .build/ bridge:

  1. Build and run Keyway (./run.sh).
  2. Chrome: chrome://extensionsDeveloper modeLoad unpackedprototype/extension/, then register the host (the ID on the extension card is fixed by the manifest's pinned key):
    cd prototype/extension
    ./install_dev_host.sh <extension-id>
  3. Firefox:
    cd prototype/extension
    ./build-firefox.sh                 # → dist-firefox/
    ./install_dev_host_firefox.sh      # no ID needed (gecko id is fixed)
    then load dist-firefox/manifest.json via about:debugging.

Unpacked (Chrome) and temporary (Firefox) extensions are dropped when the browser restarts — reload them each session. A signed Web Store / AMO build isn't published yet.

Uninstall

cd prototype
./uninstall.sh

It quits Keyway, restores Caps Lock, and removes the app plus everything it leaves behind (caches, preferences, the native-messaging host, and the permission grants). If you only have the downloaded .app, drag it to the Trash, then run hidutil property --set '{"UserKeyMapping":[]}' to restore Caps Lock if it's still acting as the trigger.

Requirements

  • macOS 13 (Ventura) or later, Apple Silicon
  • To build from source: a Swift toolchain (Xcode or the Swift CLT)
  • Two permissions, both required (granting either needs a restart to take effect — macOS caches them per process):
    • Accessibility — to read the AX tree and synthesize clicks/keys
    • Screen Recording — for the vision fallback and the settle detection

Permissions & privacy

Keyway runs entirely on your machine. No telemetry, no network calls other than the local socket between the app and the browser extension. The permissions are used only for what's described above; screen captures are processed in memory and not written to disk (outside an opt-in debug flag).

Status

Early prototype / research project. It works and is usable daily, but it is rough, unsigned, and the code lives under prototype/. Expect sharp edges. Built in the open to share the approach.

License

AGPL-3.0-or-later. Keyway bundles an icon-detection model derived from OmniParser (built on Ultralytics YOLO), whose weights are AGPL-licensed; the AGPL applies to the combined work, so the whole project is AGPL-3.0. If you run a modified version as a network service, the AGPL requires you to offer its source.

Third-party attributions are in NOTICE.md.

Acknowledgements

  • Vimium — the browser extension's element-detection heuristics are derived from it (MIT).
  • OmniParser — the on-device icon-detection model.
  • Homerow — prior art and inspiration for keyboard-driven clicking on macOS.