#AI编程
9361 related articles

Adaptive Development with Jetpack Compose: Media Query, Grid, and Nav3 in Practice
A deep dive into new Jetpack Compose adaptive development tools: Media Query capability APIs, Grid 2D layout, Flexbox, Navigation 3 scene strategies, and Compose Adaptive Skill, with large-screen adaptation benefits and case data.

Lightbuild: A Declarative Android Build Built for AI Agents
Google releases Lightbuild, an experimental Android build tool that replaces executable scripts with declarative configuration for fast IDE sync and first-class AI agent support.

Turning Cars into Canvases: A Complete Guide to Android Automotive App Development
A detailed guide to Android for Cars: Android Auto vs. Android Automotive OS, Car App Library template architecture, CAL 1.9/1.10 features, and seamless park-to-drive media transitions.

CameraX and On-Device AI: A Full Upgrade for Cross-Device Camera Experiences
Google shares the latest CameraX and on-device AI advances: Adaptive Viewfinder for foldables and XR, Feature Group API combining HDR, 60fps and stabilization, ML Kit and MediaPipe for smart cameras, and Low Light Boost coming to CameraX in Q4 2026.

Android App Security in the Agent Era: Skills, Auditing, and Runtime Protection
How Google's Android security team protects apps in the agent era: security skills close agents' context gaps, plus Play Integrity API and granular permission APIs for layered defense.

AI Agent in Practice: How WorkBuddy Lets AI Work for You
Based on a Bilibili demo video, this article analyzes WorkBuddy's WeChat assistant feature: how QR code connection lets AI execute tasks like table lookups and invoice recognition on your local computer, plus the privacy advantages and prerequisites of AI Agent adoption.

Complete Guide to Claude Cowork 2.0 Workflow Automation: Multi-Agent Collaboration in Practice
Complete guide to Claude Cowork 2.0 workflow automation. From Cowork Home folder setup and project/skill layering to scheduled tasks, connectors and browser use—build a multi-agent system step by step, with prompt ideas and pitfalls.

Building an AI Employee with Kimi Claw + Obsidian: Knowledge Base and Automation Workflow in Practice
Bilibili creator BlinkAI shares a complete tutorial on building a local AI employee with Kimi Claw + Obsidian: from web chat to delegated Agents, using Skills like Load to Daily and My Context to sink chats into a knowledge base, plus a daily automated Follow Builders task and the Claude for Kimi Obsidian sidebar chat plugin.

AI Automated Vulnerability Discovery in Practice: From Hermes Agent to Building a Workbench
Based on a Bilibili AI vulnerability discovery tutorial: from AI Agent concepts, Hermes installation and configuration, security knowledge base and Kali Linux workbench setup, to information gathering, vulnerability verification, and report generation—a complete loop, plus host security, custom Skills, and SRC practice platforms.

How to Build an Effective AI Code Review Agent: Prompt and Model Inversion in Practice
A practical breakdown of optimizing AI code review agents: from sandbox and PR webhooks to targeted prompts, historical PR calibration, model inversion, code knowledge bases, end-to-end QA, and cost tiering.

WorkBuddy in Action: How a Desktop Agent Automates Your Work
WorkBuddy is a desktop-level AI agent workbench that drives computers via command line, connects external tools, and automates file processing, data organization, and document generation. This article analyzes its working principles, complex task processing flow, and a practical path from getting started to taking orders.

Firebase AI Logic + Gemini Nano 4: A Practical Guide to Android AI Development
How Firebase AI Logic and Gemini Nano 4 help Android developers ship generative AI: App Check and prompt templates ensure security, hybrid routing cuts latency and cost, Spend Caps control budgets, and AI monitoring provides observability.

Microsoft AI Agent Breakthroughs: More Efficient, Better at Code and Weather Forecasting
Microsoft Research Focus Issue 97: ForkPilot cuts agent inference costs, Cabra reveals coding agents' code-understanding gap, Teletune learns workflows from logs, PBC improves extended forecasts, BioEMU-2 opens preview.

Prime Intellect Releases Prime Mega MoE: Trainable Fused MoE Kernel with 3.11x Forward Speedup
Prime Intellect releases Prime Mega MoE, adding backpropagation to DeepSeek's fused Mega MoE kernel for trainability. Up to 3.11x forward speedup and 34% faster Qwen3 MoE training steps in prime-rl.
Claude-Assisted Antibody Design: A New…
Claude-Assisted Antibody Design: A New Approach to Blocking IgE for Food Allergies
A study shared on Hacker News uses Claude-assisted antibody design to develop IgE-targeting molecules for food allergies. This analysis covers the approach, IgE as a target, and validation challenges.

LangChain Retry Middleware Configuration: Practices and Selection Guide
Based on Reddit community discussion, this article analyzes LangChain ModelRetryMiddleware retry parameter configuration practices, covering the selection logic and alternatives for key options like max_retries, backoff_factor, and jitter.

Vercel Flags SDK Cold Start 7x Faster, No Longer Extends Function Duration
Vercel Flags SDK optimization: cold start initialization drops from 70ms to 10ms (~7x faster); usage reporting now handled by the runtime, no longer extending function execution time. Learn how to upgrade and what it means for developers.

@ai-sdk/vue 3.0.319 Released: Routine Maintenance and Version Alignment in the AI SDK Ecosystem
@ai-sdk/vue 3.0.319 patch release syncs with core package ai@6.0.319. Here's what this version alignment means for Vue AI app developers.

phpClaw: An AI Agent Runtime Built for PHP, with Tools, Memory, and Guards
phpClaw is a PHP AI Agent runtime showcased on Hacker News, offering tool calling, memory management, and security guards, letting PHP developers implement Agent capabilities without cross-language bridging. This article covers its positioning, core capabilities, and use cases.

T3 Code: The Open-Source AI Coding Assistant's Comeback Story
T3 Code is an open-source AI coding assistant that unifies subscriptions like Codex and Claude, with powerful remote control and parallel task management.