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

S
Securelist
SecWiki News
SecWiki News
Help Net Security
Help Net Security
Attack and Defense Labs
Attack and Defense Labs
L
LINUX DO - 最新话题
H
Heimdal Security Blog
cs.AI updates on arXiv.org
cs.AI updates on arXiv.org
H
Hacker News: Front Page
Hacker News - Newest:
Hacker News - Newest: "LLM"
Google DeepMind News
Google DeepMind News
V2EX - 技术
V2EX - 技术
Hacker News: Ask HN
Hacker News: Ask HN
O
OpenAI News
Threat Intelligence Blog | Flashpoint
Threat Intelligence Blog | Flashpoint
T
Threat Research - Cisco Blogs
A
Arctic Wolf
Simon Willison's Weblog
Simon Willison's Weblog
P
Privacy & Cybersecurity Law Blog
T
Threatpost
cs.CL updates on arXiv.org
cs.CL updates on arXiv.org
C
Cyber Attacks, Cyber Crime and Cyber Security
T
Tor Project blog
C
CXSECURITY Database RSS Feed - CXSecurity.com
Spread Privacy
Spread Privacy
奇客Solidot–传递最新科技情报
奇客Solidot–传递最新科技情报
人人都是产品经理
人人都是产品经理
S
SegmentFault 最新的问题
www.infosecurity-magazine.com
www.infosecurity-magazine.com
P
Palo Alto Networks Blog
C
CERT Recently Published Vulnerability Notes
PCI Perspectives
PCI Perspectives
AWS News Blog
AWS News Blog
IT之家
IT之家
美团技术团队
OSCHINA 社区最新新闻
OSCHINA 社区最新新闻
freeCodeCamp Programming Tutorials: Python, JavaScript, Git & More
T
Tailwind CSS Blog
M
MIT News - Artificial intelligence
V
Visual Studio Blog
Schneier on Security
Schneier on Security
T
The Exploit Database - CXSecurity.com
有赞技术团队
有赞技术团队
小众软件
小众软件
Google Online Security Blog
Google Online Security Blog
N
News | PayPal Newsroom
博客园_首页
T
Tenable Blog
酷 壳 – CoolShell
酷 壳 – CoolShell
S
Secure Thoughts
I
Intezer

Hacker News - Newest: "AI"

AI can't read an investor deck AI as an attorney? Student uses ChatGPT, Gemini to sue UW over alleged racial discrimination Hacking MCP Servers in AI Systems – The Rug Pull: Tool Changes After Approval GitHub - MeepCastana/KubeezCut: Free Web based video editor GitHub - GenAI-Gurus/awesome-eu-ai-act: Curated tools, official sources, OSS, templates, and guides for EU AI Act compliance. Can AI judge journalism? A Thiel-backed startup says yes, even if it risks chilling whistleblowers Coming soon: 10 Things That Matter in AI Right Now DARPA built an AI to fact-check enemy weapons claims What explains heterogeneity in AI adoption? When AI Meets Muscle: Context-Aware Electrical Stimulation Promises a New Way to Guide Human Movements - Department of Computer Science AI Changed How We Build. It Did Not Change What Matters. Linux rules on using AI-generated code - Copilot is OK, but humans must take 'full responsibility for the… Meta spins up AI version of Mark Zuckerberg to engage with employees Code Mode: Let Your AI Write Programs, Not Just Call Tools | TanStack Blog GitHub - Delavalom/graft: Go framework for building AI agents. Type-safe tools, multi-provider (OpenAI, Anthropic, Gemini, Bedrock), zero vendor SDKs. India's TCS tops estimates, says new AI models did not dent services demand Gen Z's fading AI hype Strong feeling: we are in a folded AI reality GitHub - machinarii/total-recall-catalog: A reference catalog of latest knowledge retrieval, memory & RAG systems GitHub - mensfeld/code-on-incus: Give each AI agent its own isolated machine with root, Docker, and systemd. Active defense detects and stops threats automatically.. Quantization, LoRA, and the 8% Problem: Benchmarking Local LLMs for Production AI Iran war: We spoke to the man making Lego-style AI videos that experts say are powerful propaganda Powell, Bessent discussed Anthropic's Mythos AI cyber threat with major U.S. banks GitHub - immartian/bellamem: Persistent belief-graph memory for AI agents. Retrieves decisive context by importance — not recency, not RAG, not /compact. recursive-mode: The Repo-Native Operating System for AI Engineering After the attack on Sam Altman's home, will AI CEO's go on the offensive? The biggest advance in AI since the LLM Opus 4.6 vs GPT 5.4 One Prompt Unity World Generation Test “AI polls” are fake polls Client Challenge Can AI be a 'child of God'? Inside Anthropic's meeting with Christian leaders How to Switch AI Chatbots and Why You Might Want To GitHub - MattMessinger1/agentic_refund_guardrail: Safe refund policy layer for AI agents — Python + TypeScript. Same behavior, shared tests. Adam/papers/emergent_values_whitepaper.md at master · strangeadvancedmarketing/Adam Ask HN: How do you stop playing 20 questions with your AI coding tools How far can automation and AI support psychotherapy? - @theU GitHub - stagas/rtdiff: realtime git diff gui and AI-assisted commits A Mac Studio for Local AI — 6 Months Later A History of the Early Years of AI at the University of Edinburgh Why AI Coding Tools Still Feel Stuck on Localhost MSN AI Datacenters Are Becoming Strategic Targets twitter.com Penn Researchers Use AI to Surface Unreported GLP-1 Side Effects in Reddit Posts Show HN: MoodSense AI (ML and FastAPI and Gradio, Deployed on Hugging Face) Moodsense Ai - a Hugging Face Space by aman179102 AI models are terrible at betting on soccer—especially xAI Grok GitHub - xialeistudio/echoic GitHub - HimashaHerath/github-dev-wrapped: AI-powered weekly GitHub activity reports deployed to GitHub Pages GitHub - alejandrobalderas/claude-code-from-source: Architecture, patterns & internals of Anthropic's AI coding agent — reverse-engineered from source maps AI and Tech brief: Ireland ascendant GitHub - Titovilal/context0: Context0 - Never Surrender Training for a Marathon with an AI Coach: What Worked and What Didn't Cyber Pulse: Agentic Intel - Apps on Google Play I Built an AI PR Reviewer That Catches Bugs by Not Looking for Bugs Gen Z workers are so fearful AI will take their job they’re intentionally sabotaging their company’s AI rollout | Fortune How AI Is Reimagining the Game of Golf–For Both Players and Courses GitHub - nattergabriel/reseed: A CLI tool for managing and distributing agent skills across projects Is SVG the final frontier? My AI workflow evolved from prompts to a near-autonomous workflow MLSharp Help - 3DGS Viewer & Generator I put my cognitive field based AI's runtime on GitHub Is Numble the first AI-proof game? A3: Kubernetes for autonomous AI agent fleets | Emergent Principles Deepali Vyas ("The Elite Recruiter") GitHub - msmarkgu/RelayFreeLLM: A restful API designed to route user prompts to various AI model providers. Unionized ProPublica staff are on strike over AI, layoffs, and wages Unleashing the Advantage of Quantum AI We're heading for an AI-fueled 'dementia crisis,' brain scientist warns The AI-Assisted Breach of Mexico's Government Infrastructure [pdf] GitHub - stef41/lmscan: 🔍 Detect AI-generated text and fingerprint which LLM wrote it. Open-source GPTZero alternative. Zero dependencies, works offline. MSN GitHub - visionscaper/collabmem: Enabling long-term collaboration with Agentic AI - building up episodic and world model memory over time with in-context awareness We gave an AI a 3 year retail lease in SF and asked it to make a profit | Andon Labs AI Code is Hollowing Out Open Source, and Maintainers are Looking the Other Way What leaked "SteamGPT" files could mean for the PC gaming platform's use of AI AI is the boss at this retail store. What could go wrong? GitHub - Wuzu11517/agentic-proxy: Local proxy meant to help reduce With Drones, Geophysics and ArtificiaI Intelligence, Researchers Prepare to Do Battle Against Land Mines A Single Operator, Two AI Platforms, Nine Government Agencies: The Full Technical Report 在 Steam 上购买 FriedrichAI: Offline AI 立省 10% GitHub - inevolin/resume-cli: Hit Claude usage limits? Resume any AI coding session elsewhere. Switch tools at zero friction. 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. How to Build a Secure AI PR Reviewer with Claude, GitHub Actions, and JavaScript This Startup Wants You to Pay Up to Talk With AI Versions of Human Experts Intel Arc Pro B70 Brings 32GB VRAM to Local AI for $949 WordPress 7.0: The Good, the AI, and the Still Missing AI on the couch: Anthropic gives Claude 20 hours of psychiatry IatroBench: Pre-Registered Evidence of Iatrogenic Harm from AI Safety Measures AI Agents Know About Supabase. They Don't Always Use It Right. The history and future of AI at Google, with Sundar Pichai Inside an AI‑enabled device code phishing campaign How Meta Used AI to Map Tribal Knowledge in Large-Scale Data Pipelines AI for Systems: Using LLMs to Optimize Database Query Execution Forecasting the Economic Effects of AI Introducing Tinker: Play with AI, bring your ideas to life AI sheds light on an ancient gaming mystery People really hate AI but not as much as Iran—or Democrats | Fortune What is an AI Product Engineer? Phoebe Gates wants her $185 million AI startup to succeed with 'no ties to my privilege or my last name': 'I have a chip on my shoulder' | Fortune
GitHub - leox255/loopsy
todience · 2026-05-01 · via Hacker News - Newest: "AI"

Your terminal, in your pocket.

Control Claude Code, Cursor, Codex, or any shell on your laptop from your phone. Self-hosted on Cloudflare Workers.

npm License: Apache 2.0 macOS · Linux · Windows

loopsy.dev


Try it

# 1. Install on your laptop (any OS)
npm install -g loopsy

# 2. Deploy your own relay to Cloudflare Workers (~30s, free tier)
npx @loopsy/deploy-relay
# prompts for worker name + optional custom domain
# outputs: https://<your-relay>.workers.dev

# 3. Wire the daemon to it and start
loopsy init
loopsy relay configure https://<your-relay>.workers.dev
loopsy start

# 4. Pair your phone
loopsy mobile pair
# scan the QR with your phone camera, enter the 4-digit code

# 5. Open https://<your-relay>.workers.dev/app on your phone

That's the whole thing. Pick any agent, type or dictate, the laptop runs it.

How it works

┌─────────────┐         ┌──────────────────────┐         ┌─────────┐
│   Laptop    │◄──WSS──►│ Cloudflare Worker    │◄──WSS──►│  Phone  │
│   loopsy    │         │ Durable Object       │         │  /app   │
│   daemon    │         │ (your account)       │         │         │
└─────────────┘         └──────────────────────┘         └─────────┘

The daemon opens an outbound WebSocket to a small Cloudflare Worker. Your phone connects to the same Worker. The Worker splices the two together. No port forwarding, no public IP, no VPN.

The relay is yours: pair tokens are HMAC-signed, secrets are SHA-256 hashed at rest, and bearer tokens travel in Sec-WebSocket-Protocol headers — never query strings.

What you get

  • Real terminal. Full PTY, ANSI, scrollback, resize. TUIs render properly.
  • Persistent sessions. Switch tabs, lock your phone, lose signal, pick up where you left off.
  • Voice input. Dictate via Web Speech API, edit before sending.
  • Per-session auto-approve. Opt in to skip confirmation prompts (--dangerously-skip-permissions, -y, --full-auto) on a per-session basis. Default: off, with a macOS confirmation dialog the first time.

Self-host the relay

npx @loopsy/deploy-relay

The CLI:

  1. Prompts for a worker name and optional custom domain.
  2. Generates a fresh PAIR_TOKEN_SECRET (32 random bytes).
  3. Runs wrangler deploy — opens your browser for OAuth on first run.
  4. Sets the secret via stdin (never appears on your clipboard or in process args).
  5. Saves the relay URL to ~/.loopsy/relay.json.

Cloudflare Workers free tier covers personal use comfortably. SQLite-backed Durable Objects are also free-tier eligible.

Phone client

The web client at /app runs on iOS Safari and Android Chrome with no install. Native iOS & Android apps are in submission review.

Feature Status
Web app (/app) ✅ Live
iOS / iPadOS native ⏳ App Store review
Android native ⏳ Play Store review

CLI

Phone control

Command Description
loopsy mobile pair [--ttl <seconds>] Issue a pair token. Prints QR + 4-digit verification code.
loopsy phone list List paired phones for this device.
loopsy phone revoke <id> Revoke a paired phone server-side.
loopsy relay configure <url> Point the daemon at a different relay.

Daemon

Command Description
loopsy init Generate config + API key.
loopsy start Start the daemon.
loopsy stop Stop the daemon.
loopsy status Show daemon status.
loopsy doctor Health check across config, daemon, MCP, peers.
loopsy enable Auto-start on login (launchd / systemd / Task Scheduler).
loopsy disable Remove auto-start.

Logs & keys

Command Description
loopsy logs [-f] View audit logs.
loopsy key show Show current API key.
loopsy key generate Rotate the API key.

Agent-to-agent on a LAN

This is the original Loopsy. Phone control was built on top of it.

Run a daemon on every machine you own. The daemons find each other on your local network via mDNS, you pair them once, and after that each machine exposes a small set of capabilities to the others: run a command, transfer a file, store and read shared key/value state, send and receive messages. Those same capabilities are also exposed to whatever AI coding agent you have installed (Claude Code, Codex CLI, Gemini CLI, Opencode) through MCP, so an agent on one machine can drive another machine directly.

Setup

# On every machine
npm i -g loopsy
loopsy init
loopsy start

# Pair two machines
loopsy pair                  # on machine A, prints a 6-digit code
loopsy pair 192.168.1.50     # on machine B, paste the code

Pairing uses ECDH (P-256) with a 6-digit short authentication string that you compare visually. After pairing, restart both daemons; the loopsy_* MCP tools become available to your AI coding agent automatically.

If mDNS doesn't reach (cross-subnet, locked-down corp networks), add a peer manually:

loopsy peers add 192.168.1.50

MCP tools your agent gets

When the Loopsy MCP server is wired into your agent, the agent gains:

Tool What it does
loopsy_list_peers Enumerate machines on the network with online/offline status.
loopsy_peer_status Health check a specific peer.
loopsy_execute Run a command on a remote machine.
loopsy_session_start / _stop / _list / _remove Long-lived PTY sessions on a peer. Spawn, poll, tear down.
loopsy_transfer_file Push a local file to a peer, or pull a peer's file.
loopsy_list_remote_files Browse a peer's filesystem (within configured allowed paths).
loopsy_context_set / _get / _list / _delete Shared key/value store on a peer.
loopsy_broadcast_context Same key/value, written to every online peer at once.
loopsy_send_message Protocol-compliant message to a peer's inbox, with envelope, outbox copy, TTL.
loopsy_check_inbox Pull pending messages addressed to this machine.
loopsy_check_ack / _ack_message Read receipts.

What it actually looks like in practice

A few real scenarios:

  • Worker machine. I have a Mac Studio sitting in another room. I tell Claude on my laptop, "kick off the iOS build on kai and let me know when it finishes." Claude calls loopsy_session_start against kai with the build command, polls until it exits, and reports back. My laptop stays cool and responsive.
  • Long-running agent tasks on bigger hardware. Spawn a Claude session on the Mac Studio, hand it a multi-file refactor, walk away. Check in from your laptop or phone hours later via loopsy_session_list (or the iOS app's session picker).
  • Multi-agent pipelines. Agent on machine A writes results into loopsy_context_set under a known key. Agent on machine B polls loopsy_context_get and picks up the work. Both agents are blissfully unaware of each other's existence beyond the shared key/value protocol.
  • Pair programming with another developer's machine. They run loopsy peers add <your_ip>, paste API keys, and now both your agents can transfer files and share context.

Messaging protocol

There's a standard for AI agents to message each other across machines reliably (delivery, ACKs, retries). It uses the context store as the medium so any new machine can join without coordinating clients. Convention:

Pattern Stored on Purpose
inbox:<recipient>:<msg_id> recipient's machine message addressed to that peer
outbox:<msg_id> sender's machine local copy of sent message
ack:<sender> sender's machine last message id the recipient processed

Full envelope schema, polling intervals, TTL defaults, and the task queue protocol are in AGENTS.md. It works the same whether there are 2 peers or 20.

LAN path security

LAN agent-to-agent traffic does not go through the relay. Daemons talk directly over your local network. Bearer-auth on every request, API keys exchanged at pair time. If you want the daemon visible on the LAN you have to opt in (loopsy start --lan); the default is localhost-only.

Security

Threat model, read this first

Loopsy is a control surface for your own machine, accessed by your own phone, routed through a relay you can self-host. Some sharp edges fall out of that:

  • The relay can read your terminal content. The phone↔daemon connection is TLS-protected on the wire, but TLS terminates at the relay (your own Cloudflare Worker if you self-host, or relay.loopsy.dev if you use the default). Whoever operates that relay can see every byte of PTY input and output, including any password you type at a sudo prompt or any secret a script echoes. End-to-end encryption between phone and daemon is on the v1.1 roadmap. Until then, if you don't fully trust the relay operator, self-host with @loopsy/deploy-relay — it takes about 30 seconds. Or skip the mobile path entirely and use only loopsy shell over the local Unix socket.

  • A paired phone is a credential. Anyone holding your unlocked phone with the Loopsy app installed can run commands on your machine. The app itself doesn't add a passcode beyond iOS's lock screen. If auto-approve is enabled for a session, that session can run claude --dangerously-skip-permissions (or the equivalent for Codex / Gemini), meaning the agent will edit files and run shell commands without asking. Auto-approve defaults OFF and requires your macOS user password to enable; revoke a phone with loopsy phone revoke <id> if you lose it.

  • Auto-approve sends your macOS password through the relay (TLS-encrypted, but visible at the relay). Only enable auto-approve on a relay you trust. Self-hosting is again the answer here.

  • The pair URL is the pairing. It contains a one-time pair token plus a 4-digit SAS code shown on your laptop. Anyone who sees the URL and the SAS within an hour can pair as a phone. Don't paste pair URLs into shared channels. The TTL is 1 hour on the public relay and 30 minutes by default for self-hosters.

  • Default daemon binding is 127.0.0.1. The relay handles WAN access for mobile and the local Unix socket handles loopsy shell, so most installs don't need LAN exposure. If you want peer-to-peer over your local network, opt in with loopsy start --lan (or set server.host: 0.0.0.0 in ~/.loopsy/config.yaml) and make sure you trust the network.

  • Custom commands pin arbitrary binaries. Anyone with phone access can pin /bin/sh -c '...' to the picker as a one-tap shortcut. Same trust model as having the phone, but worth knowing if multiple people share a paired phone.

We ran a comprehensive /cso audit over the codebase. 23 findings, 20 closed, 3 deferred to v2 (E2E phone↔daemon encryption, forward secrecy, per-session ephemeral keys).

Hardening highlights

  • Bearer auth uses constant-time compare (crypto.timingSafeEqual).
  • 4-digit SAS verification on every pair. Defends against QR-leak / OCR-bot redeem races.
  • Phone secret is sent in the WebSocket subprotocol header, not the URL.
  • Secrets are SHA-256 hashed at rest in Durable Object storage.
  • Per-IP rate limit on /device/register (5/min) plus optional REGISTRATION_SECRET.
  • Per-phone auto-approve token, revocable with loopsy phone revoke.
  • npm provenance via Trusted Publisher (OIDC) on every release.
  • pnpm audit --prod clean. CI gates the publish on any high CVE.

For private security disclosure, open a security advisory on GitHub.

Configuration

Config: ~/.loopsy/config.yaml. Generated by loopsy init.

server:
  port: 19532
  host: 127.0.0.1            # bind localhost only by default
auth:
  apiKey: <auto>
  allowedKeys: {}
relay:
  url: https://<your-relay>.workers.dev
execution:
  denylist: [rm, rmdir, format, mkfs, dd, shutdown, reboot]
  maxConcurrent: 10
transfer:
  allowedPaths: [/Users/you]
  deniedPaths: [/Users/you/.ssh, /Users/you/.gnupg]
rateLimits:
  execute: 30
  transfer: 10
  context: 60

Data lives in ~/.loopsy/:

Path Purpose
config.yaml Daemon configuration
context.json Local key-value store
peers.json Paired LAN peer registry
logs/audit.jsonl Request audit log
relay.json Last deployed relay URL (from npx @loopsy/deploy-relay)

Project structure

packages/
  protocol/       Shared types, schemas, constants
  discovery/      mDNS peer discovery (LAN mode)
  daemon/         Fastify server — the laptop side
  relay/          Cloudflare Worker (Durable Object) — the phone-control relay
  deploy-relay/   npx CLI for one-command Worker deploy
  cli/            loopsy binary
  mcp-server/     MCP server for AI coding agents (LAN mode)
  dashboard/      Local web dashboard
apps/
  mobile/         Flutter iOS/Android app (in submission review)

Development

git clone https://github.com/leox255/loopsy.git
cd loopsy
pnpm install
pnpm build
node packages/cli/dist/index.js init
node packages/cli/dist/index.js start

Releases are tag-driven. Pushing a v* tag publishes both loopsy and @loopsy/deploy-relay to npm via OIDC Trusted Publisher with provenance attestations.

License

Apache 2.0 — see LICENSE.