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

I
InfoQ
G
Google Developers Blog
Engineering at Meta
Engineering at Meta
月光博客
月光博客
博客园 - 聂微东
博客园 - Franky
奇客Solidot–传递最新科技情报
奇客Solidot–传递最新科技情报
有赞技术团队
有赞技术团队
A
About on SuperTechFans
Microsoft Azure Blog
Microsoft Azure Blog
Blog — PlanetScale
Blog — PlanetScale
U
Unit 42
T
Tailwind CSS Blog
Cyber Security Advisories - MS-ISAC
Cyber Security Advisories - MS-ISAC
云风的 BLOG
云风的 BLOG
S
SegmentFault 最新的问题
F
Fortinet All Blogs
H
Help Net Security
J
Java Code Geeks
酷 壳 – CoolShell
酷 壳 – CoolShell
博客园 - 叶小钗
L
LangChain Blog
Martin Fowler
Martin Fowler
N
Netflix TechBlog - Medium

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 - secondly-com/OpenPhone: AI-first open source ope...
talboren · 2026-06-26 · via Hacker News: Show HN

OpenPhone GitHub hero

CI Release License Status

OpenPhone is an AI-native Android OS that turns the phone into an agentic device: a system-level AI agent that can see the screen, operate apps, remember commitments, monitor phone events, and continue work in the background with user review and auditability built into the OS.

This repository is the canonical OpenPhone entry point. It contains the OpenPhone-owned Android overlay, privileged assistant app, framework patches, model/tool policy configuration, build scripts, device notes, contracts, and release tooling. It intentionally does not vendor the full Android source tree.

AI-Native Phone Runtime

OpenPhone is built around a system-level agent, not a chatbot app. The agent is installed as a privileged OS component with an always-available system surface for conversation, realtime voice, approvals, active runs, and proactive state. Actions are mediated through OpenPhone framework services instead of brittle app-layer automation.

The agent can read structured phone context and use model-visible tools to work across apps. Context includes the foreground app, visible UI hierarchy, screen text and controls, notifications, calls, messages, calendar state, location, battery, connectivity, active watchers, background runs, and commitments the user made in conversation. Sensitive actions are reviewable, and behavior can be inspected through audit logs, trajectories, screenshots, policy decisions, and release validators.

OpenPhone is also built for proactive work. Heartbeats quietly check whether anything needs attention. Scheduled jobs run exact workflows. Watchers monitor phone context such as missed calls, messages, notifications, foreground app state, visible screen state, calendar changes, location, battery, connectivity, and commitments the user made in conversation. Background runs keep working after the current chat turn, while the system surface shows what is running, why it started, what it last said, and what needs review.

The current developer preview is based on LineageOS 23.2 / Android 16 and targets Google Pixel 9a (tegu) first.

Use Cases

  • "Catch me up on everything important from overnight" - consume missed calls, messages, notifications, calendar changes, and reminders, then return a short morning gist.
  • "Order me an Uber to the office" - open the right app, set the destination, select a ride, and stop for review before booking.
  • "Play something random on Spotify" - open Spotify, choose music, and continue until playback actually starts.
  • "If I miss a call from this number, send them 'I'll call you back soon'" - create a watcher tied to future call context and message policy.
  • "Watch for delivery updates and only bother me if something changes" - turn notification noise into a targeted background monitor.
  • "Help me finish this screen" - inspect the visible app state, identify the next control, and act through OS-mediated taps or text input.
  • "Remind me when this conversation becomes relevant" - turn a commitment into durable state that can resurface later based on time, app, or phone context.
  • "Keep working on this after I leave" - continue a multi-step task as a visible background run with approval where needed.

Repository Layout

.github/       CI, release, eval, contribution, security, issue, and PR files.
docs/          Product docs, device notes, legal docs, releases, and testing.
manifests/     Android repo local manifests.
overlay/       OpenPhone-owned files copied into the Android tree.
patches/       Patch stacks applied on top of upstream LineageOS repos.
schemas/       Machine-readable runtime contracts and release/eval schemas.
scripts/       Sync, patch, build, flash, validation, and release helpers.
services/      Reference services, including the development model broker.

Start with docs/README.md if you are looking for a specific document.

How It Works

flowchart TB
  User["User<br/>voice, touch, text, volume chord"]
  Surface["OpenPhone system surface<br/>chat, realtime voice, approvals, runs"]
  Assistant["Privileged OpenPhoneAssistant<br/>orchestrator, model adapters, tool loop"]
  Runtime["Proactive runtime<br/>heartbeats, scheduled jobs, watchers, background runs"]
  Context["Phone context<br/>visible UI, foreground app, notifications, calls, messages,<br/>calendar, location, battery, commitments"]
  Services["OpenPhone OS services<br/>openphone_agent, openphone_context, openphone_assistant_data"]
  Policy["Policy, approval, and evidence<br/>capabilities, confirmations, audit logs, trajectories"]
  Android["Android framework<br/>ActivityTaskManager, WindowManager, InputManager, NotificationManager"]
  Apps["Apps and device<br/>Settings, Messages, Spotify, Uber, Pixel 9a"]

  User --> Surface
  Surface --> Assistant
  Assistant --> Runtime
  Assistant --> Services
  Runtime --> Services
  Services --> Context
  Services --> Policy
  Services --> Android
  Android --> Apps
  Apps --> Context
  Policy --> Surface
Loading

The high-level architecture is documented in docs/ARCHITECTURE.md. The capability model is in docs/CAPABILITIES.md, and machine-readable contracts live under schemas.

Quick Start

Validate the repository:

./scripts/check.sh
git diff --check

Install repo if needed:

./scripts/install-repo.sh

Sync and patch the Android tree:

./scripts/sync.sh
./scripts/apply-patches.sh

Build the generic OpenPhone ARM64 product for validation:

./scripts/build.sh openphone_arm64

Build the Pixel 9a target on a Linux Android build host:

OPENPHONE_BUILD_GOAL="droid target-files-package otapackage" \
  ./scripts/build.sh openphone_tegu

For assistant-only iteration on an already flashed development device:

OPENPHONE_BUILD_GOAL=OpenPhoneAssistant ./scripts/build.sh openphone_tegu
scripts/push-assistant-apk.sh /path/to/OpenPhoneAssistant.apk

Full build instructions are in docs/BUILD.md. Testing and physical eval guidance is in docs/TESTING.md.

Device Support

OpenPhone builds on LineageOS device infrastructure, so the broader universe of potential ports starts with the official LineageOS supported-device list: wiki.lineageos.org/devices. OpenPhone support is narrower: a device is only supported after it has an OpenPhone product target, flash/recovery notes, hardware validation, agent validation, and release coverage.

The first OpenPhone physical target is Google Pixel 9a (tegu). Generic ARM64 builds are useful for product graph validation, but they are not a supported phone target.

OpenPhone does not redistribute Google apps, Google Mobile Services, vendor blobs, signing keys, private firmware, or restricted device material. Local developer GMS sideload notes are in docs/GMS.md.

See docs/devices/MATRIX.md and docs/devices/tegu.md.

Community

Contributions, issues, and device validation reports are welcome under the terms in .github/CONTRIBUTING.md.

Commercial Use

OpenPhone-owned materials are source-available for non-commercial use under the PolyForm Noncommercial License 1.0.0. Commercial use requires a separate written license from Dafdef, inc.

Contributions are accepted only under terms that allow Dafdef, inc. to own, modify, sublicense, redistribute, and commercialize the submitted work. See .github/CONTRIBUTING.md, docs/legal/COMMERCIAL.md, LICENSE, docs/legal/LICENSE.noncommercial, and docs/legal/THIRD_PARTY_NOTICES.md.