AI 教程与实战
收录 AI 工具教程、提示词、安装部署、配置指南、开发实战与可复现步骤。
分类文章8440
当前页43 / 211
持续更新永久归档
分类文章
按发布时间从新到旧 · 每页 40 条
教程 / 实战普通
Preparing Product Data and Pages for AI Shopping Discovery
AI-powered shopping experiences are adding a new layer to product discovery. Rather than presenting a...
信息来源:DEV Community
教程 / 实战普通
MCP Token Cost: I Cut My MCP Server From 15 Tools to 4
I rebuilt a real AWS cost-triage MCP server from 15 tools to 4. Same account, same question, 7 tool calls down to 2, schemas 88% smaller.
信息来源:DEV Community
教程 / 实战普通
Keeping track of your agent's work
Long-running agents lose context, and so do you. Why every feature or fix I work on with an agent gets a ticket the agent keeps up to date.
信息来源:DEV Community
教程 / 实战普通
Agentic Search May Be the Bigger Opportunity Than GEO, but Measurement Comes First
Generative Engine Optimization is useful shorthand for improving a site’s chances of being...
信息来源:DEV Community
教程 / 实战普通
My agent has two Playwrights
1,929 Playwright MCP calls in a month and almost none were tests. How I use the browser as my agent's eyes, and why every PR now ships with screenshots.
信息来源:DEV Community
教程 / 实战普通
Seven months of coding agents: what $28,335 actually bought
201,473 tool calls across five agent harnesses, and 0.7% of them ran while I slept. What the spend, the interrupts and the 15-actions-per-turn leash say about staying in the loop.
信息来源:DEV Community
教程 / 实战普通
MCP 2.0: the release that deleted the handshake
MCP's 2026-07-28 revision removed sessions, the initialize handshake and held-open streams. What the handshake era cost, what replaced it, and why Appwrite's MCP server shipped it in a day.
信息来源:DEV Community
教程 / 实战普通
Will MCP replace the CLI?
MCP gives agents typed, discoverable tools, but the CLI remains reproducible and composable. Where each interface wins and why both will survive.
信息来源:DEV Community
教程 / 实战普通
Connect your Garmin to Claude in two minutes
A step-by-step guide to adding a free Garmin connector to Claude on the web, phone and desktop, what it can see, and the first questions to ask. Part 1 of Your Garmin data in Claude.
信息来源:DEV Community · Claude
教程 / 实战普通
AI Says My Project Will Take 2 Months. Then Builds It in a Week.
I noticed something strange about AI project planning. Ask AI to create a development plan for a...
信息来源:DEV Community
教程 / 实战普通
What Happens When Someone Tricks Your AI Agent Into Ignoring Its Rules?
You spent three weeks building an AI agent that handles customer refunds. It validates receipts,...
信息来源:DEV Community
教程 / 实战普通
MCP — Model Context Protocol
MCP — Model Context Protocol A deep-dive walkthrough of connecting AI applications to...
信息来源:DEV Community
教程 / 实战普通
What should stop an AI coding agent before it acts?
Agent safety discussions often collapse three different questions: Role preflight — Is this task...
信息来源:DEV Community
教程 / 实战普通
Playwright docs say dispatch_event('click') equals element.click(). On a disabled button it didn't.
Four ways to click from a Playwright script, run against a disabled, a hidden and an off-screen button. On the disabled one, dispatch_event ran the onclick and element.click() did not.
信息来源:DEV Community
教程 / 实战普通
Your secrets don't need a git commit to leak into an AI tool
Most of us have the "don't commit secrets" reflex drilled in. Pre-commit hooks, secret scanners in...
信息来源:DEV Community
教程 / 实战普通
How I make motion videos with Claude Opus 5.5 (the full setup, prompts and numbers)
TL;DR. Opus 5.5 never touches a pixel. It writes one HTML file where every frame is a pure function...
信息来源:DEV Community · Claude
教程 / 实战普通
Who left this note for your AI agent? (A 2-minute check for Codex and Claude Code users)
A web page can save an AGENTS.md into your Downloads folder without a single click. Your coding agent...
信息来源:DEV Community · Claude
教程 / 实战普通
How to Measure AI Search Visibility Beyond Rankings and Clicks
AI-generated search answers make familiar SEO reporting incomplete. A brand can be named in an answer...
信息来源:DEV Community
教程 / 实战普通
Skillware Still 6 Months Ahead of Every Major Agent Skills Framework
Why the AI agent ecosystem is hitting a wall with ad-hoc tool calling, and how Skillware's...
信息来源:DEV Community
教程 / 实战普通
How a Key Leaked in May Shipped Malware in October
The LiteLLM attack had no clever trick in it. Every step was a key or token that still worked long after anyone needed it. And the malware ran before Python imported anything.
信息来源:DEV Community
教程 / 实战普通
Automated API Load Testing on AWS: Grafana k6 and AI-Powered Reporting with Amazon Bedrock
1. Introduction When developing an application or service, one of the common questions we...
信息来源:DEV Community
教程 / 实战普通
Top 5 Open Source LLM Gateways
When an application uses AI models from multiple providers, having a good open source LLM gateway can...
信息来源:DEV Community
教程 / 实战普通
I Coded for 30 Days Using Only Claude Code — Here's What I Learned
Not "AI-assisted." Not "copilot suggestions I edited." I made a rule: for 30 days I don't type...
信息来源:DEV Community · Claude
教程 / 实战普通
Context Engineering over Prompt Engineering: Architecting Codebases for AI Agents
Context Engineering over Prompt Engineering: Architecting Codebases for AI Agents TLDR:...
信息来源:DEV Community
教程 / 实战普通
Your AI remembers the old answer. How do you handle facts that changed?
I work at Wontopos, so I'm biased. I build a memory API. But this post isn't about it. It's about a...
信息来源:DEV Community
教程 / 实战普通
Black-Box Testing with AI Agents: Test the Rules, Not the Code
One of the most useful test reports I've seen an AI produce came from agents that never saw a single...
信息来源:DEV Community
教程 / 实战普通
Yaron Minsky 演讲要点:如何改进代码库以适配 AI 智能体
Lauren Tan 转述 Yaron Minsky 关于为智能体改进代码库的演讲要点:智能体依赖反馈,类型是最好的反馈之一,用类型让非法状态不可表示可在全代码库强制执行全局规则。
信息来源:X:Lauren Tan(SpaceXSI,Grok Bot 工程负责人) (@poteto) · Grok
教程 / 实战普通
Pira - Creating my own Jira in Code Apps
I've always worked by a golden rule for AI generated software, the 'Blast Zone must be small', what...
信息来源:DEV Community
教程 / 实战普通
The LiteLLM Master Key Leak Sat Unrevoked for 136 Days, Then Hijacked a Release Pipeline
The LiteLLM master key leak sat unrevoked for 136 days after a force-push supposedly fixed it, then let an attacker publish a malicious PyPI release.
信息来源:DEV Community
教程 / 实战普通
I stopped hunting for good first issues. A Gemma inside GitHub Actions does it while I sleep
This is a submission for the Hacktoberfest Open-Source AI Challenge Week 3: Slow and Steady ...
信息来源:DEV Community
教程 / 实战普通
Claude Code 如何做 Code Review:用分层检查与快速回滚替代逐行人工审阅
作者引述 Anthropic 工程实践,介绍 Claude Code 团队在做约九成代码由 Claude 编写、变更量暴涨的背景下如何重做 Code Review。
信息来源:X:邵猛 (@shao__meng) · Claude
教程 / 实战普通
OpenAI 工程师如何用 Codex 的 Ultrafast 模式把事故调查从数小时缩短到几分钟
OpenAI 工程师已把 Codex 的高速模式 Ultrafast 当作事故响应的默认工具,可同时排查多个潜在原因。安全调查从过去的一两小时缩短到 10 至 15 分钟,多数回答在 30 秒内返回。工程师还用它做代码重构和浏览器表单填写等日常任务。
信息来源:OpenAI:官网动态(RSS · 排除企业/客户案例) · OpenAI
教程 / 实战普通
SlothAssist: An On-Device Gemma Agent for Automated Shopping
This is a submission for the Hacktoberfest Open-Source AI Challenge: Week 3 — Slow and Steady. ...
信息来源:DEV Community
教程 / 实战普通
Sloth Migrate
This is a submission for the Hacktoberfest Open-Source AI Challenge Week 3: Slow and Steady Remember...
信息来源:DEV Community
教程 / 实战普通
Build without limits - SDLC Briefing, #3
Hi! I am Ricardo Sueiras, leading SDLC transformation at Indicium AI. That means I get to work with...
信息来源:DEV Community
教程 / 实战普通
Gemma Reads the Chat. Python Refuses to Guess.
This is a submission for the Hacktoberfest Open-Source AI Challenge Week 3: Slow and Steady * ...
信息来源:DEV Community
教程 / 实战普通
SlothPace 🦥: Taming Chaotic Morning Brain Dumps with 100% Offline Open-Source AI
This is a submission for the Hacktoberfest Open-Source AI Challenge Week 3: Slow and Steady ...
信息来源:DEV Community
教程 / 实战普通
LeakLens: A Swiss-Editorial Financial Autopilot Powered by TabPFN-3.5 That Audits Messy Bank CSVs
This is a submission for the Hacktoberfest Open-Source AI Challenge: Week 3 ("Slow and...
信息来源:DEV Community
教程 / 实战普通
Claude Code 团队如何做代码评审
Claude Code 团队介绍其代码评审流程:Claude 写了约 90% 的代码,6 月到 9 月合并的 PR 数量翻倍以上,需要回退的 PR 占比下降超过 40%。
信息来源:X:Addy Osmani (@addyosmani) · Claude
教程 / 实战普通
Day closed. You can stop carrying it now. SlothCloser, a local evening wind-down agent.
This is a submission for the Hacktoberfest Open-Source AI Challenge Week 3: Slow and Steady ...
信息来源:DEV Community