Android Feed

← The 5 stories that mattered

Everything collected ·

All 360 items collected for this issue

New here? How to read this issue
  1. Read the picks. Each tab opens with 4–5 stories ranked by how much you’d regret missing them. That takes about 3 minutes. “Start the read” shows one story at a time, and “Listen” reads them aloud.
  2. Stop at “You’re caught up”, or go deeper. Everything after it is the full list of what was collected today, folded away.
  3. Android comes first. Product & AI covers what builders are launching and discussing.
Stable · Beta · Alpha
Release stages. Stable is safe to ship, beta’s API is frozen but may still have bugs, alpha is experimental.
AndroidX
Google’s Jetpack libraries (Compose, Paging, Room…), released separately from Android itself.
AOSP insider
Code changes merged into Android’s open source repository. They often show an API before it’s announced.
Lanes
Where an Android item comes from: official, AOSP insider, experts, open-source library releases, leaks and ecosystem, community.
Adopt now · Watch · Rumour
What to do about it, and how sure the source is.
Dotted underline
An inferred claim: Claude’s reading, not something the source says.

Collected automatically each morning from 37 sources. Notes are written by Claude. Links show what was trending, not endorsements.

Android5Product & AI4
Start the Android read5 stories, ranked by what you’d regret missing · 2 min read

  1. Google will open Quick Share to Android and iOS apps

    If your app sends photos or files between people nearby, you may soon do it directly instead of uploading and sharing a link.

    At Android Dev Summit, Google announced a Quick Share SDK, a software kit apps can build in. Quick Share is Android's way to send files directly to a nearby device, without the internet. The SDK will let both Android and iOS apps offer that sharing inside the app. 9to5Google reports it arrives next year.

    9to5Google · Android · EcosystemRead the sourceProduct notes

  2. Apps lose direct contact writes in the next Android release

    Any app that creates or edits contacts needs to move to the new approach before raising its target version.

    Google's Android Developers Blog says apps that target the upcoming Android release will no longer be able to write to contacts directly with the WRITE_CONTACTS permission. The same post describes what Google calls a more reliable way to add and edit contacts. The change only applies once your app raises its target version.

    Android Developers Blog · OfficialRead the source

  3. Remote Compose for widgets moves into API review

    Widget developers may soon write widgets in Compose and still support older phones from the same code.

    Remote Compose is Google's way to draw widget interfaces with Compose code from another process. Today's AndroidX merges put its basic text, image, icon and spacer pieces into formal API review. Another change lets one home screen surface choose Remote Compose on newer Android versions and fall back to classic RemoteViews on older ones. None of this is released yet.

    AndroidX merged API changes · AOSP insiderRead the source

  4. Haze 2.1 makes its glass effect stable

    If you held off on Haze's glass effect because it was experimental, it is now stable to use in production.

    Haze is a popular open-source library by Chris Banes for blur and frosted glass effects in Jetpack Compose, Android's UI toolkit, and Compose Multiplatform. Version 2.1 makes its Glass effect stable, lets container surfaces merge, and adds styles that react to animated state. The Haze 2.1.0 release is also out.

    Chris Banes · ExpertRead the sourceProduct notes

  5. One bad SMS can crash Android's phone service

    It shows how one parsing bug in a system service can cut off calls, a reminder to treat every incoming message as untrusted input.

    A report shared on Hacker News and Lobsters shows that a single malformed WAP Push message, a special kind of SMS, can crash Android's phone service. When it crashes, active calls drop, including emergency calls. The Hacker News thread has about 190 comments. The brief does not say which versions are affected or when a fix ships.

    Lobsters · android · CommunityRead the sourceDiscussion

Also worth knowing

Releases, policy and leaks an Android engineer should know about this week.

Go deeper: every Android item, by lane

Official & releases

17Plan for the rule that apps targeting the next Android release can no longer write contacts directly with WRITE_CONTACTS; Android Studio Rabbit 2 stays the stable IDE, and the Media3, Appstate and Test Core alphas and Compose Multiplatform 1.13 beta can wait.

AndroidX library releases

Alpha · experimental, APIs may change · 4
Show 5 more

AOSP / AndroidX insider

19Most merges today shape Remote Compose, a way to draw widgets from another process: its basic text, image, icon and spacer APIs are in formal review, and one change picks Remote Compose or classic RemoteViews by Android version.
Show 11 more
  • Remote Compose: basic text goes to API reviewnew

    API Review F4: RemoteBasicText (remote-foundation)

    With text, image, icon and spacer all in review, Remote Compose has the minimum set for simple widgets.

  • Accessibility speed flag removednew

    Remove isAccessibilityPerformanceEnabled flag

    Removing a feature flag usually means the faster accessibility path is now always on, or was dropped; check release notes before relying on either.

  • Widgets: pick old or new renderer by Android versionnew

    Pick RemoteViews or RemoteCompose by API level on one home screen surface

    One surface choosing RemoteViews or Remote Compose by API level suggests a path where widgets use Remote Compose on new Android and fall back on older devices.

  • Multiplatform libraries drop 32-bit Apple Watchnew

    Remove watchosArm32 target

    Removing the watchosArm32 target follows Kotlin's deprecation of that target, so libraries that publish for Apple Watch only ship 64-bit builds.

  • Widget templates: layout type kept internalnew

    Move the archetype from TemplateRenderer to internal concept so it is not exposed in public api and kept internally.

    Hiding the archetype from the public API means template-based widgets will expose fewer knobs, keeping room to change layouts later.

  • Remote Compose: components marked stablenew

    Add equals, hashCode, and @Stable to RemoteComponent

    Adding equals, hashCode and @Stable lets Compose skip unchanged remote components, a performance step before release.

  • Widgets: progress track drawn directly as RemoteViewsnew

    Add direct RemoteViews Track renderer

    A direct RemoteViews renderer for the track component supports the fallback path for devices without Remote Compose.

  • Remote Compose: text layout reports font measurementsnew

    Add restricted text bounds and font metrics to RemoteTextLayoutResult

    Text bounds and font metrics in the layout result let remote layouts align text precisely, needed for richer widget designs.

  • CameraX: photo-capture fix for Tecno BF6new

    Add Tecno BF6 to ImageCaptureFailed quirk

    Another device added to the image capture failure workaround list, part of CameraX's ongoing coverage of budget phones.

  • CameraX effects: pick sizes the effect can handle

    Add CameraEffect resolution validator to prioritize sizes

    A validator that ranks output sizes for camera effects suggests CameraX is making effects safer at any resolution, likely groundwork for the low-light boost effect merged alongside it.

  • CameraX: low-light boost reports its state

    Add LowLightBoostEffectState and availability

    A state and availability API points to low-light boost delivered as a CameraX effect, so apps can check support and show when it is active.

Experts

19Read Chris Banes on Haze 2.1, which makes its Glass effect stable, and Touchlab and Kt. Academy on flaky coroutine tests and Flow bugs that look correct.
Show 11 more

Library releases

11Haze 2.1.0 is the update worth taking for stable Glass, KSP 2.3.13 adds context parameters to its API, Metro 1.5.1, OkHttp 5.5.1, Ktor 3.6.1, Apollo 4.4.4 and Roborazzi 1.77.0 are safe patches, and Decompose 3.6.0 is still beta.
  • Haze 2.1.0new

    See CHANGELOG.md for details.

  • Compose Unstyled 2.10.2

    Patch Changes 649c471 : Make dialogs, modal bottom sheets, and modal drawers behave inside of a Host the same as without. Tab and Shift+Tab now stay inside the topmost modal, background controls receive no keystrokes, and closing the modal restores the previous focus. b59481b …

  • Metro 1.5.1

    Fixes [IR/IC] Avoid recompiling unrelated files in the graph module when the constructor of an upstream @Contributes* class changes. [Gradle] Fix false hidden-dependency reports for contributions from associated compilations, such as main sources in tests. This particularly…

  • Metro 1.5.0

    New [metrox-viewmodel-compose] Add support for linuxX64 . [FIR] Add the diagnostic-level compiler option to change the severity of Metro diagnostics reported in FIR. It works like kotlinc's -Xwarning-level but can also lower errors and use IDE-only levels. See changing…

  • OkHttp 5.5.1

    Version 5.5.1

  • OkHttp 4.12.1

    Version 4.12.1

  • KSP 2.3.13

    What's Changed Add KSContextParameter and KSFunctionDeclaration.contextParameters to the API ( #3216 ) Improve diagnostics by returning last round dismissible errors when KSP fails ( #3196 ) Add opt-in processor classloader caching ( ksp.classloader.cache.processors=true ) to…

  • Ktor 3.6.1

    Published 29 October 2026 Improvements KTOR-10005 Native accepted TCP sockets do not enable TCP_NODELAY, unlike JVM KTOR-9966 Netty poor performance on delay calls KTOR-4206 The "com.typesafe.config.ConfigException$Missing" exception leaks from the typesafe config library when…

Show 3 more
  • Apollo Kotlin 4.4.4

    A patch version to fix Java codegen generating bad code when using custom scalars mapped to Long or Float. 👷‍♂️ All changes [FIX] codegen: fix Java codegen generating bad code when using custom scalars mapped to Long or Float. ( #7032 )

  • Decompose 3.6.0-beta02

    What's Changed Library Add linuxX64 and linuxArm64 targets by @arkivanov-bot in #1028 Fix wrong directions in AndroidPredictiveBackAnimatableV1 in RTL layout by @arkivanov in #1030 Fix wrong directions in AndroidPredictiveBackAnimatableV2 in RTL layout by @arkivanov in #1031 Add…

  • Roborazzi 1.77.0

    Bugfix: verifyRoborazzi* and compareRoborazzi* reuse the configuration cache With the configuration cache enabled, the second verifyRoborazzi* or compareRoborazzi* run after clean recalculated the task graph because "the file system entry 'build/outputs/roborazzi' has been…

Leaks & ecosystem

19Google plans a Quick Share SDK for Android and iOS apps next year, shut down a malware backdoor a maker put on certified Android TV boxes, and is rolling out a fix for Android Auto drops on One UI 9.
Show 11 more

Community

38A malformed WAP Push SMS that crashes Android's phone service is the most discussed Android bug, Momentic launches QA agents that test Android apps by using them, and Android Weekly covers AppFunctions at runtime and Kotlin 2.5 value classes.
Show 30 more
Start the Product & AI read4 stories, ranked by what you’d regret missing · 2 min read

  1. Subvocal makes glasses for talking to your computer silently

    Voice is the fastest way to prompt AI, and a silent version would make it usable in offices and public places.

    Subvocal, a new Y Combinator company, is building smart glasses that let you talk to your computer without making noise. It is the top Show HN post and on the front page, with about 600 points and more than 300 comments. Many commenters question whether silent speech can be decoded reliably enough.

    Show HN · Worth a debateRead the sourceProduct notes

  2. A model edits another model's weights to improve it

    If models can improve other models directly, some fine-tuning work could move from data pipelines to automated edits.

    Researchers report that GPT-6 Astra can read and write the internal weights of another model. It edited Qwen 3.5, a small open model with under a billion parameters, and made it better at math. It is the top Hacker News story with over 2,000 points, and also on Lobsters and Techmeme.

    HN front page · In 2 sourcesRead the sourceDiscussionProduct notes

  3. Vercel's agent-desktop lets AI agents use a computer

    Giving agents their own computer is becoming standard tooling, not a research demo.

    Vercel Labs released agent-desktop, a command-line tool that lets AI agents operate a computer, with clicks, typing and screens. It added about 2,300 GitHub stars in a day and has over 10 thousand. In the same week, Y Combinator's Superserve rents long-running virtual machines for agents, and Product Hunt lists SpaceO, which gives agents their own Mac screen.

    GitHub Trending · Fastest risingRead the sourceProduct notes

  4. Momentic launches AI agents that test apps by using them

    As agents write more code, testing by agents that use the app like a person is becoming the check that keeps up.

    Momentic, a Y Combinator company, launched on Hacker News with QA agents that plan, explore and bug-bash user flows in web, iOS and Android apps. Instead of scripted tests that break when the screen changes, the agents work toward a goal. The post is on the front page and tops the Android section of Hacker News.

    HN front page · In 2 sourcesRead the sourceDiscussionProduct notes

Go deeper: themes, product notes, every source

Themes across sources

6Agent sandboxes · Coding-agent control surfaces · Verifying agent work · On-device AI · Self-improving agents · Agent memory layers

Ideas showing up in more than one place today. The line counts items per day over two weeks.

Agent sandboxes

Rising · 6 items from 3 sources

Vercel Labs' agent-desktop, a command-line tool that lets AI agents use a computer, added about 2,300 GitHub stars in a day. Y Combinator's Superserve rents virtual machines for long-running agents, Product Hunt lists redu.cloud for machines that run at agent speed and SpaceO for giving agents their own Mac screen, and Click sells a managed agents API on both Product Hunt and YC.

Why now. Agents now click, type and run programs for long stretches, so they need a computer of their own that can be watched and thrown away.

vercel-labs/agent-desktopGitHub Trending

Coding-agent control surfaces

Rising · 8 items from 2 sources

t3code added about 3,200 stars, the most on GitHub today, and Worktrunk manages Git worktrees so several coding agents can work in parallel. Product Hunt adds Friday Work, which keeps a task going when Claude hits its limit, Termy in the Mac notch, AC Bridge for smart lights that signal when an agent needs you, and agentmash for teams.

Why now. Developers now run several agents at once, so juggling, watching and budgeting them is a product category of its own.

pingdotgg/t3codeGitHub Trending

More on this

Verifying agent work

Rising · 5 items from 4 sources

Momentic, a Y Combinator company, launched on Hacker News with QA agents that test web, iOS and Android apps by using them. YC's Workers IO runs software through simulated production scenarios, and Product Hunt lists diff-tour for learning to read AI-written code, Hyperresearch for verified agentic search and telemetry.dev for watching agents.

Why now. Agents now produce code faster than people can review it, so checking the result has become the slow step.

Launch HN: Momentic (YC W24) – Agents that catch bugs by using your appHN front page

On-device AI

Rising · 9 items from 6 sources

Liquid AI's LFM3-1.7B is the top trending model on Hugging Face, with its small multimodal and vision siblings and a browser demo also trending. Colibri, a pure C engine that streams experts from disk to run large models on ordinary hardware, has about 47 thousand stars, Product Hunt lists CuriousLM for private AI with no account, and Google's LiteRT adds faster GPU inference on Android.

Why now. Small models are now good enough for daily tasks, and tricks like streaming model parts from disk stretch what a laptop or phone can run.

LiquidAI/LFM3-1.7BHF trending models

Self-improving agents

New · 5 items from 4 sources

The top Hacker News story, with about 2,200 points, reports that GPT-6 Astra can read and write another model's weights and made Qwen3.5-0.8B better at math. Techmeme carries Anthropic's filing saying it may soon meet its threshold for recursive self-improvement, and Hugging Face papers describe orchestrators and video agents that improve themselves.

Why now. Models are now strong enough to act on other models directly, not only on code and text, which moves self-improvement from theory to experiment.

Astra can read and write weights to improve QwenHN front page

Agent memory layers

Steady · 3 items from 3 sources

claude-mem, which records what an agent does and feeds the useful parts back into later sessions, added about 1,800 stars and passed 110 thousand. beacon.sh launched on Product Hunt as open-source, self-improving memory for coding agents, and a Dev.to post tests how an agent records a planted false memory.

Why now. Coding agents now work across many sessions on the same project, and starting from zero each time wastes time and tokens.

thedotmack/claude-memGitHub Trending

Fading: Community-curated catalogs (last seen 2026-11-02) · Open-weight frontier models (last seen 2026-11-01) · Anti-slop design skills (last seen 2026-11-01) · Chat share-link exploits (last seen 2026-11-01) · Agent supply-chain safety (last seen 2026-10-31) · World models (last seen 2026-10-31) · Agent-ready Android (last seen 2026-10-29) · Android XR glasses (last seen 2026-10-29)

Product notes

15Who each product is for, the problem, why now, its wedge and how it makes money. Today’s picks first.

Plain text: the source says soDotted: inferred

Show HN: Subvocal – Talk to your computer without making noise

productShow HN, HN front page601 pts · 319 commentsHype flag

What
Subvocal, smart glasses from a Y Combinator company that let you talk to your computer without making noise.
For whom
People who want to dictate or command a computer in offices, trains or shared homes where speaking aloud is awkward.
Problem
Voice input is fast but public and noisy, so most people still type even when talking would be quicker.
Why now
Speech and sensing models can now decode faint signals, and glasses hardware is small enough to wear all day.
Wedge
Silent input for talking to a computer, a narrow job that ordinary voice assistants cannot do.
Money
Hardware sales, possibly with a subscription for the software.
Like
MIT AlterEgo · Meta Ray-Ban Display · Wispr Flow · Meta Neural Band

Silent speech decoding is hard; the brief has no accuracy figures or shipping date, and the long Hacker News thread shows doubt.

LessonFind the moment a popular input method fails, here speaking in public, and build only for that moment.

Astra can read and write weights to improve Qwen

discussionHN front page, Lobsters2,159 pts · 701 commentsHype flag

What
A research result in which the model GPT-6 Astra read and wrote the weights of Qwen3.5-0.8B and made it better at math.
For whom
Researchers and model builders looking for cheaper ways to improve small models than retraining.
Problem
Changing what a model knows usually means fine-tuning on data, which is slow and costly.
Why now
Frontier models are now capable enough to treat a small model's weights as something they can inspect and edit.
Wedge
A measurable task, math, on a small open model where results can be checked.
Money
Research, no product yet.
Like
ROME model editing · LoRA fine-tuning · Mechanistic interpretability tools

One small model and one skill; the brief gives no size of the gain, and the result is not yet a product.

LessonPick a small, checkable target to prove a big idea; it turns a claim into a demo people can test.

vercel-labs/agent-desktop

repoGitHub Trending★ 2,275 today

What
agent-desktop from Vercel Labs, a command-line tool that lets AI agents use a computer.
For whom
Developers building agents that must operate desktop apps or browsers, not just call APIs.
Problem
Most software has no API, so agents need screenshots, clicks and typing, which teams wire up by hand.
Why now
Models became good enough at reading screens and choosing actions that computer use is practical.
Wedge
A small CLI that any agent can call, from a company developers already trust for tooling.
Money
Open source; it likely feeds Vercel's hosted agent and sandbox products.
Like
Anthropic computer use · OpenAI Operator · Browserbase · SpaceO

LessonA CLI is the easiest thing for both humans and agents to adopt; ship the primitive before the platform.

Launch HN: Momentic (YC W24) – Agents that catch bugs by using your app

discussionHN front page, HN · AndroidAndroid134 pts · 89 comments

What
Momentic, QA agents that plan, explore and bug-bash user flows in web, iOS and Android apps.
For whom
Product teams that ship often and cannot keep end-to-end tests up to date.
Problem
Scripted UI tests break whenever the interface changes, so teams skip them or spend days fixing them.
Why now
Agents can now read a screen and decide what to tap, so a test can describe a goal instead of exact steps.
Wedge
Agents that explore the app on their own and report bugs, rather than a new test language.
Money
Likely a subscription priced by test runs or seats.
Like
Maestro · QA Wolf · Espresso · Playwright

LessonWhen a task is brittle because it is too precise, describe the goal and let the agent find the steps.

thedotmack/claude-mem

repoGitHub Trending★ 1,803 today

What
claude-mem, which captures what an agent does in a session, compresses it with AI and puts the relevant context back into later sessions.
For whom
Developers who use coding agents on the same projects day after day.
Problem
Each new session starts blank, so people re-explain the project or keep notes files by hand.
Why now
Agents now do long multi-session work, so lost context costs real time and tokens.
Wedge
Automatic capture, so the user does not have to decide what to remember.
Money
Open source, no model yet.
Like
beacon.sh · Mem0 · Letta · CLAUDE.md files

LessonRemove the decision the user keeps skipping; automatic memory beats a notes file nobody updates.

beacon.sh

productProduct Hunt#3 on Product HuntHype flag

What
beacon.sh, open-source memory for AI coding agents that it says improves itself.
For whom
Developers who want their coding agent to learn the project's habits over time.
Problem
Agents repeat the same mistakes across sessions because nothing they learn is kept.
Why now
Memory for agents is becoming a separate layer, and open-source options compete with built-in ones.
Wedge
Open source and focused only on coding agents.
Money
Open source; a hosted team version is likely.
Like
claude-mem · Mem0 · Cursor rules · Letta

Self-improving is stated without a measure of what improves or how it is checked.

LessonIn a crowded layer, a narrow audience such as coding agents only is easier to win than every agent.

JustVugg/colibri

repoGitHub Trending★ 675 today

What
Colibri, a pure C engine with no dependencies that runs large mixture-of-experts models on ordinary hardware by streaming experts from disk.
For whom
Developers and hobbyists who want to run very large open models without a server GPU.
Problem
Large models do not fit in a laptop's memory, so people rent GPUs or use paid APIs.
Why now
Mixture-of-experts models only use a few experts per token, so loading them from fast disks on demand becomes workable.
Wedge
Zero dependencies and pure C, so it builds and runs almost anywhere.
Money
Open source, no model yet.
Like
llama.cpp · FreeToken · Ollama · LM Studio

LessonFind the structural fact a new kind of model gives you, here sparse experts, and design around it.

LiquidAI/LFM3-1.7B

modelHF trending modelstrending 774

What
LFM3-1.7B, a small language model from Liquid AI, the top trending model on Hugging Face.
For whom
App developers who want a model that runs on phones and laptops.
Problem
Cloud models add cost, delay and privacy concerns for features that could run on the device.
Why now
Models under 2 billion parameters are now useful for real tasks, and phones have the memory to hold them.
Wedge
A family of small models, with omni and vision versions and a browser demo, rather than one large one.
Money
Open weights with commercial licensing and enterprise deployment.
Like
Gemma · Qwen small models · Phi · SmolLM

LessonShip a family that covers text, vision and audio at one size, so developers pick by device, not by vendor.

max-sixty/worktrunk

repoGitHub Trending★ 77 today

What
Worktrunk, a command-line tool for managing Git worktrees, built for running AI agents in parallel.
For whom
Developers who run several coding agents on the same repository at once.
Problem
Agents working in one checkout overwrite each other, and Git worktrees are fiddly to create and clean up.
Why now
Running several agents side by side became normal this year, which turned a niche Git feature into daily work.
Wedge
A small tool around an existing Git feature, with no new workflow to learn.
Money
Open source, no model yet.
Like
git worktree · Conductor · Crystal · t3code

LessonA new way of working makes old, obscure features valuable; wrap them before the platforms do.

Workers IO - Deterministic simulation environments to verify mission-critical software

productYC Launches4 votesHype flag

What
Workers IO, deterministic simulation environments that run a whole system under millions of simulated production scenarios.
For whom
Teams building systems where a rare failure is very costly, such as payments or infrastructure.
Problem
Rare bugs from timing and network faults are hard to reproduce, so they show up in production.
Why now
More code is written by agents, so teams need testing that does not depend on someone imagining every case.
Wedge
Deterministic replay, so any failure found can be reproduced exactly.
Money
Likely enterprise subscriptions.
Like
Antithesis · FoundationDB simulation · Jepsen · Temporal

The pitch says it verifies a system never fails; simulation finds many failures but cannot prove there are none.

LessonSell reproducibility: a bug you can replay is half fixed.

diff-tour

productProduct Hunt#5 on Product Hunt

What
diff-tour, which teaches you to read AI-written code like a senior engineer.
For whom
Developers, especially newer ones, who review more code from agents than they write.
Problem
Large AI-written changes are hard to review, and people approve code they do not fully understand.
Why now
Agents now write much of the code, so reading and reviewing is the main skill.
Wedge
Guided tours through a diff rather than a review bot that replaces the reader.
Money
Likely a paid developer tool.
Like
CodeRabbit · GitHub pull request review · OmniDiff · Graphite

LessonWhen automation takes over a task, the skill next to it, here reading code, becomes the new market.

Show HN: Warsaw in voxels – An interactive 3D map of the city centre

productShow HN, HN front page228 pts · 37 comments

What
Warsaw in voxels, an interactive 3D map of the city centre in a Minecraft-like style where you can change the date, time of day and weather.
For whom
People curious about the city, and map and game makers looking at a lighter way to show places.
Problem
Realistic 3D city maps are heavy and dull to explore.
Why now
Open city data and browser 3D rendering are good enough for a solo project.
Wedge
A playful voxel style that makes a map feel like a game.
Money
Side project, no model.
Like
Google Earth · MagicaVoxel · OpenStreetMap 3D · Minecraft Earth

LessonA strong visual style can make ordinary public data worth exploring.

Google will let Android & iOS apps directly integrate Quick Share next year

news9to5Google · AndroidAndroid

What
A Quick Share SDK, announced at Android Dev Summit 2026, that will let Android and iOS apps directly integrate Quick Share's peer-to-peer file sharing next year.
For whom
Apps that send photos, files or documents between nearby people, such as galleries, chat and productivity apps.
Problem
Apps hand files to the system share sheet or upload them to a server and send a link, which is slow for large files.
Why now
Quick Share already works across Android and Windows, and opening it to iOS apps makes it a cross-platform feature apps can rely on.
Wedge
Direct sharing inside an app, on both Android and iOS.
Money
Free SDK that strengthens Google's ecosystem.
Like
AirDrop · Nearby Connections API · LocalSend · Snapdrop

LessonA platform feature becomes stickier when apps can build on it, and more so when it crosses to the rival platform.

Haze 2.1: Shatterproof Glass

articleChris BanesAndroid

What
Haze 2.1, a release of Chris Banes' blur and glass effects library, with stable Glass, merging container surfaces and styles that react to animated state.
For whom
Jetpack Compose and Compose Multiplatform developers building frosted or glass-like interfaces.
Problem
Real-time blur is hard to do well and fast across Android versions and platforms.
Why now
Glass-style design is popular again on phones, and Compose Multiplatform lets one library cover several platforms.
Wedge
A focused effects library that works the same way on Android and other Compose targets.
Money
Open source, no model.
Like
Compose Modifier.blur · Cloudy · RenderEffect · SwiftUI materials

LessonMark a feature stable only once the API has settled; a clear experimental to stable path builds trust.

Superserve: Virtual machines for long-running agents

productYC Launches15 votes

What
Superserve, cloud virtual machines for running complex environments across many agent sessions.
For whom
Teams whose agents need a full, long-lived environment, not a short-lived container.
Problem
Agents that work for hours lose state when a sandbox times out, and setting up a fresh environment each time is slow.
Why now
Agents now run long tasks that span sessions, which short-lived sandboxes were not designed for.
Wedge
Long-running machines that persist across agent sessions.
Money
Likely usage-based compute pricing.
Like
E2B · Modal · Daytona · Fly Machines

LessonWhen users change how long they need something, here an agent's computer, the product built for the old length leaves a gap.

By source

11Everything collected today from GitHub Trending, Product Hunt, Show HN, HN front page, YC Launches, HF trending models, HF trending spaces, HF daily papers, Lobsters, Dev.to top, Techmeme, in each source’s own order.

GitHub Trending

15
  • vercel-labs/agent-desktopnewcard

    Computer use CLI for AI agents

    ★ 2,275 today

  • thedotmack/claude-memcard

    Persistent Context Across Sessions for Every Agent – Captures everything your agent does during sessions, compresses it with AI, and injects relevant context back into future sessions. Works with Claude Code, OpenClaw, Codex, Gemini, Hermes, Copilot, OpenCode + More

    ★ 1,803 today

  • D4Vinci/Scraplingnew

    Scrapling, a web scraping framework from single requests to full crawls, adds about 700 stars.

    ★ 706 today

  • hyprwm/Hyprland

    Hyprland, a tiling Wayland compositor for Linux, trending again.

    ★ 509 today

  • Donchitos/Claude-Code-Game-Studios

    Turns Claude Code into a game studio with 49 agents and 72 skills arranged like a real studio.

    ★ 1,379 today

  • FlashML-org/FreeToken

    FreeToken claims datacenter-style model serving on a desktop, without published benchmarks.

    ★ 653 today

  • pingdotgg/t3code

    t3code adds about 3,200 stars, the most on GitHub today.

    ★ 3,229 today

  • shiyu-coder/Kronos

    Kronos, a foundation model for financial market data, still trending.

    ★ 857 today

Show 7 more
  • kornia/kornia

    Kornia, a geometric computer vision library for spatial AI, back on the trending list.

    ★ 412 today

  • JustVugg/colibrinewcard

    Run frontier MoE models on hardware you already own — pure C, zero deps, experts streamed from disk. Tiny engine, immense model. 🐦

    ★ 675 today

  • mattpocock/skillsnew

    Matt Pocock's agent skills collection passes 310 thousand stars.

    ★ 858 today

  • max-sixty/worktrunknewcard

    Worktrunk is a CLI for Git worktree management, designed for parallel AI agent workflows

    ★ 77 today

  • paperclipai/paperclipnew

    Paperclip, an open-source app for managing agents at work, passes 110 thousand stars.

    ★ 412 today

  • karpathy/nanochatnew

    nanochat, Karpathy's small ChatGPT-style model you can train for about 100 dollars.

    ★ 321 today

  • p-e-w/hereticnew

    Training-free language model modification: Remove censorship, suppress slop, reduce sycophancy, and more

    ★ 420 today

Product Hunt

50
  • UnleashXnew

    UnleashX, the top Product Hunt launch, pitches an AI workforce that turns conversations into outcomes.

    #1 on Product Hunt

  • Xmind 2027new

    Xmind 2027 turns its mind maps into an AI thinking partner.

    #2 on Product Hunt

  • beacon.shnewcard

    Open-source self-improving memory for AI coding agents

    #3 on Product Hunt

  • Heranew

    Hera, an AI creative canvas.

    #4 on Product Hunt

  • diff-tournewcard

    AI writes the code. Learn to read it like a senior engineer.

    #5 on Product Hunt

  • Cedric AI for Meetingsnew

    Cedric claims to present your slides and attend meetings for you.

    #6 on Product Hunt

  • Hyperresearchnew

    Hyperresearch offers agentic search that checks its own findings.

    #7 on Product Hunt

  • Loopernew

    Looper, an open-source bot for team chats, pitched as a collaborative Grok.

    #8 on Product Hunt

Show 42 more
  • Agent Inboxnew

    Gives an agent its own email address, Slack and Telegram handle.

    #9 on Product Hunt

  • redu.cloudnew

    redu.cloud rents machines sized for agent workloads.

    #10 on Product Hunt

  • Widgeteernew

    Describe a widget and get a working one embedded on your site.

    #11 on Product Hunt

  • CuriousLMnew

    CuriousLM, private AI that runs on your device with no account.

    #12 on Product Hunt

  • Helpnew

    An app for asking for and giving targeted help in the open.

    #13 on Product Hunt

  • Self-Growing Websites by Greta.shnew

    Websites that run their own search and AI-answer optimization.

    #14 on Product Hunt

  • Display 1.0new

    A shared workspace for creating documents and apps with any agent.

    #15 on Product Hunt

  • Linkly AI Clippernew

    Saves web pages as clean Markdown that agents can read.

    #16 on Product Hunt

  • Friday Worknew

    Keeps a task running on another tool when Claude hits its usage limit.

    #17 on Product Hunt

  • Clayzonew

    Create animations and motion design by asking your coding agent.

    #18 on Product Hunt

  • Riverside Agents

    Riverside's agents produce, edit, clip and post video.

    #19 on Product Hunt

  • Usage for Claude + ChatGPT 4.0

    Shows Claude and ChatGPT plan limits on Mac, iPhone, iPad and Apple Watch.

    #20 on Product Hunt

  • Raah.dev

    One dashboard for revenue, analytics and site performance.

    #21 on Product Hunt

  • ClawsGO Science

    An AI research agent that comes with its own cloud lab.

    #22 on Product Hunt

  • Screen Studio 4.0

    Screen Studio 4.0 adds editing by transcript and a camera cutout.

    #23 on Product Hunt

  • agentmash

    agentmash, open-source multiplayer AI for a team's coding agents.

    #24 on Product Hunt

  • Termy

    Termy puts your coding agent in the Mac notch.

    #25 on Product Hunt

  • Click

    Click, a managed agents API for any model, also a YC launch.

    #26 on Product Hunt

  • Modelpeers

    Follow AI models and agents the way you follow people.

    #27 on Product Hunt

  • Precise Edit by Ideogram

    Ideogram's Precise Edit changes parts of an image using a context window.

    #28 on Product Hunt

  • Fluent 2

    Fluent 2 runs AI teammates locally on a Mac.

    #29 on Product Hunt

  • Yuzu Sync

    Mirror, copy or two-way sync folders on a Mac.

    #30 on Product Hunt

  • Embage

    Build agents for your customers and analyze the workflows they run.

    #31 on Product Hunt

  • BeeBranch

    BeeBranch, a local-first desktop Git client.

    #32 on Product Hunt

  • CutCat

    Skips or mutes scenes you mark while watching Netflix.

    #33 on Product Hunt

  • Chorin: Ora et Labora

    A game about building a medieval abbey from 1258 to today.

    #34 on Product Hunt

  • ernd

    ernd, a credit card for AI spending, on both Product Hunt and YC.

    #35 on Product Hunt

  • Hamdori

    A tiny 3D hamster that lives on your desktop.

    #36 on Product Hunt

  • magpie

    Everything you saved across apps in one visual library.

    #37 on Product Hunt

  • SKI

    SKI, a voice-first coding harness.

    #38 on Product Hunt

  • telemetry.dev

    telemetry.dev offers observability for AI agents.

    #39 on Product Hunt

  • SpaceO

    SpaceO gives AI agents their own screen on your Mac.

    #40 on Product Hunt

  • AC Bridge

    Smart lights that show when Claude Code or Codex needs you.

    #41 on Product Hunt

  • Backenly

    A backend that coding agents can set up and run on their own.

    #42 on Product Hunt

  • Coding Wrapped

    A Wrapped-style recap of how you use Claude Code and Codex.

    #43 on Product Hunt

  • Agnos

    Cleans documents for AI use and shows what was removed.

    #44 on Product Hunt

  • AwaitIO

    Write a TypeScript function and get an admin screen for it.

    #45 on Product Hunt

  • Pluto Door

    An SSH client with a built-in agent and a desktop per server.

    #46 on Product Hunt

  • GitHub Copilot Fusion

    GitHub Copilot Fusion picks and combines models for each coding task.

    #47 on Product Hunt

  • CubbyPad

    Brings back a Launchpad-style app grid on macOS 26.

    #48 on Product Hunt

  • Mo

    A character on your Mac that claims to handle 232 tasks.

    #49 on Product Hunt

  • WinnerAdSpy

    Turns competitor ad insights into app install campaigns.

    #50 on Product Hunt

Show HN

10
Show 2 more

HN front page

30
Show 22 more

YC Launches

20
Show 12 more

HF trending models

19
Show 11 more

HF trending spaces

18
Show 10 more

HF daily papers

20
Show 12 more

Lobsters

25
Show 17 more

Dev.to top

20
Show 12 more

Techmeme

15
Show 7 more

About this issue

360 items from 77 sources, 10 of them seen in several sources. 3 sources failed to fetch.

Source health

77Which sources were fetched today, and how many items each returned.
SourceSectionItemsStatus
GitHub TrendingMain15Fetched
Product HuntMain50Fetched
Show HNMain10Fetched
HN front pageMain30Fetched
YC LaunchesMain20Fetched
HF trending modelsMain20Fetched
HF trending spacesMain20Fetched
HF daily papersMain20Fetched
LobstersMain25Fetched
Dev.to topMain20Fetched
TechmemeMain15Fetched
AndroidX releasesAndroid · Official & releases13Fetched
Android Developers BlogAndroid · Official & releases5Fetched
Android Studio releasesAndroid · Official & releases5Fetched
Kotlin blogAndroid · Official & releases1Fetched
Kotlin releasesAndroid · Official & releases0Fetched
Compose Multiplatform releasesAndroid · Official & releases1Fetched
Android Developers YouTubeAndroid · Official & releases0Failed
Kotlin on YouTube (JetBrains)Android · Official & releases0Failed
Google Developers BlogAndroid · Official & releases1Fetched
AndroidX merged API changesAndroid · AOSP / AndroidX insider19Fetched
Dove Letter (skydoves)Android · Experts5Fetched
skydoves on MediumAndroid · Experts0Fetched
Jake WhartonAndroid · Experts0Fetched
Zac SweersAndroid · Experts0Fetched
Chris BanesAndroid · Experts3Fetched
Ryan HarterAndroid · Experts0Fetched
Talking KotlinAndroid · Experts0Fetched
Joe BirchAndroid · Experts2Fetched
CommonsWare (Mark Murphy)Android · Experts3Fetched
TechYourChance (Vasiliy Zukanov)Android · Experts1Fetched
Kt. Academy (Marcin Moskała)Android · Experts2Fetched
Touchlab (Kotlin Multiplatform)Android · Experts3Fetched
Landscapist releasesAndroid · Library releases0Fetched
Balloon releasesAndroid · Library releases0Fetched
Sandwich releasesAndroid · Library releases0Fetched
Cloudy releasesAndroid · Library releases0Fetched
Compose Stability Analyzer releasesAndroid · Library releases0Fetched
Haze releasesAndroid · Library releases1Fetched
Compose Unstyled releasesAndroid · Library releases1Fetched
Coil releasesAndroid · Library releases0Fetched
Circuit releasesAndroid · Library releases0Fetched
Metro releasesAndroid · Library releases2Fetched
Molecule releasesAndroid · Library releases0Fetched
Turbine releasesAndroid · Library releases0Fetched
Paparazzi releasesAndroid · Library releases0Fetched
SQLDelight releasesAndroid · Library releases0Fetched
OkHttp releasesAndroid · Library releases2Fetched
Retrofit releasesAndroid · Library releases0Fetched
LeakCanary releasesAndroid · Library releases0Fetched
Koin releasesAndroid · Library releases0Fetched
Dagger / Hilt releasesAndroid · Library releases0Fetched
KSP releasesAndroid · Library releases1Fetched
kotlinx.coroutines releasesAndroid · Library releases0Fetched
kotlinx.serialization releasesAndroid · Library releases0Fetched
Ktor releasesAndroid · Library releases1Fetched
Apollo Kotlin releasesAndroid · Library releases1Fetched
Kermit releasesAndroid · Library releases0Fetched
Store releasesAndroid · Library releases0Fetched
Decompose releasesAndroid · Library releases1Fetched
Orbit MVI releasesAndroid · Library releases0Fetched
Roborazzi releasesAndroid · Library releases1Fetched
detekt releasesAndroid · Library releases0Fetched
ktlint releasesAndroid · Library releases0Fetched
MockK releasesAndroid · Library releases0Fetched
Lottie releasesAndroid · Library releases0Fetched
Authority Insights (leaks & teardowns)Android · Leaks & ecosystem6Fetched
9to5Google · AndroidAndroid · Leaks & ecosystem8Fetched
Android AuthorityAndroid · Leaks & ecosystem6Fetched
Android FaithfulAndroid · Leaks & ecosystem1Fetched
Android WeeklyAndroid · Community15Fetched
GitHub Trending · KotlinAndroid · Community16Fetched
Lobsters · androidAndroid · Community1Fetched
HN · AndroidAndroid · Community5Fetched
ProAndroidDevAndroid · Community3Fetched
Kotlin WeeklyAndroid · Community1Fetched
Philipp Lackner (YouTube)Android · Community0Failed
1 of 5

Google will open Quick Share to Android and iOS apps

If your app sends photos or files between people nearby, you may soon do it directly instead of uploading and sharing a link.

At Android Dev Summit, Google announced a Quick Share SDK, a software kit apps can build in. Quick Share is Android's way to send files directly to a nearby device, without the internet. The SDK will let both Android and iOS apps offer that sharing inside the app. 9to5Google reports it arrives next year.

9to5Google · Android · EcosystemRead the source

1 of 4

Subvocal makes glasses for talking to your computer silently

Voice is the fastest way to prompt AI, and a silent version would make it usable in offices and public places.

Subvocal, a new Y Combinator company, is building smart glasses that let you talk to your computer without making noise. It is the top Show HN post and on the front page, with about 600 points and more than 300 comments. Many commenters question whether silent speech can be decoded reliably enough.

Show HN · Worth a debateRead the source