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5606
当前页83 / 141
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
Why I built an open-source alternative to Microsoft's System Center Orchestrator (Windows Automation)
Why I started building NodePilot, a modern alternative to System Center Orchestrator I...
信息来源:DEV Community
教程 / 实战普通
Capbroker: I gave an AI agent a fake GitHub key, then watched it get tricked into trying to delete a repo anyway
Most AI-agent setups today hand the agent the real API key directly — in an environment variable, a...
信息来源:DEV Community
教程 / 实战普通
You.com Adds Background Research Tasks for Asynchronous API and n8n Workflows
You.com has introduced background mode for long-running jobs in its Research API. Instead of keeping...
信息来源:DEV Community
教程 / 实战普通
The Sycophancy Problem: Why AI Keeps Agreeing With You (And Why That's Actually a Bug)
We have all experienced it before. Ask an AI assistant if your business idea is good. Ask it if your...
信息来源:DEV Community
教程 / 实战普通
Your Prompt Didn't Change. Your Model Did. Now What?
Every few months the same thing happens. A provider ships a new model. The one you pinned gets a...
信息来源:DEV Community
教程 / 实战普通
How I built an AI Coding Mentor (KODA) entirely on a $150 Android phone 📱🐯
Most people think you need a $2,000 MacBook and a CS degree to build an AI-powered developer tool....
信息来源:DEV Community
教程 / 实战普通
Optimizing for the 8GB Barrier: Strategic Model Selection for Local AI
Navigating the 8GB Hardware Landscape in 2026 For most developers, the dream of running a...
信息来源:DEV Community
教程 / 实战普通
Debug Log #3: I Had Claude Code Backtest a Trading Strategy for a Week, and It Found Three Bugs In Its Own Simulation
The number said "-50%." I'd been staring at trade logs long enough to have a rough feel for how far a...
信息来源:DEV Community · Claude
教程 / 实战普通
The LLM can declare that a task "appears to be finished. It will not be possible to check it off.
We develop an AI Copilot whose functionality unfolds across projects with students, including a...
信息来源:DEV Community
教程 / 实战普通
Cooley 如何用 ChatGPT Work 加速 IPO 业务
国际律所 Cooley 基于 ChatGPT Work 打造自研 AI 产品 GO Public,用智能体框架分析 IPO 所需的海量信息,让律师把精力集中在需要判断力的环节。Cooley 在 2025 年参与 180 笔交易,交易总额超 515 亿美元。该产品让团队从客户自身信息出发,结合公开来源与精选先例,更快形成定制化的 IPO 起点。
信息来源:OpenAI:官网动态(RSS · 排除企业/客户案例) · OpenAI / ChatGPT
教程 / 实战普通
"The agent said it would wait, and didn't"
"Two agents told me they were going to wait, and that's where they stopped; ten minutes later the harness killed them, with not a single ticket finished and USD 2.79 spent. The fix was one line of code — and five days later it still lives …
信息来源:DEV Community
教程 / 实战普通
Adecco Group 在 40 多个国家向 27,000 名员工部署 Salesforce Agentforce Coworker
Adecco Group 在英法试点后,于 9 月 15 日宣布在 40 多个国家向 27,000 名员工铺开 Salesforce 的 Agentforce Coworker,该助手基于 Anthropic 的 Claude,支持销售与招聘工作。
信息来源:Artificial Intelligence News(网页) · Claude
教程 / 实战普通
I had a model translate my locale file. The bug it introduced was correct Japanese.
430 strings, four models, four languages. Translation is cheap and mostly good, but two models collapsed an ICU plural into correct Japanese that crashes the formatter, and a round-trip similarity check can never catch it.
信息来源:DEV Community
教程 / 实战普通
AI Can Write the Code. Can It Prove the Fix?
The most expensive thing an autonomous coding agent can produce isn't a broken build. A broken build...
信息来源:DEV Community
教程 / 实战普通
How Do You Actually Test an AI System? A Layered Strategy From Five Tools I Built
Testing AI is not testing code, so I built five open-source tools around one idea: you cannot assert on a single right answer, you can only ask whether it got worse.
信息来源:DEV Community
教程 / 实战普通
Your Agent Burns LLM Money on Switch Statements. Jev Claims 444x Less
In TypeSafe's own evals, Jev ties Claude Sonnet 5 at 67.8% for about 1/294th the cost per case. Move your agent's decisions off the LLM, in TypeScript.
信息来源:DEV Community · Claude
教程 / 实战普通
I built a real-time flood digital twin of a UK river that runs entirely in the browser, GPU and on-device AI included
I've been chasing GPU cycles since machines had 64K of RAM, and every few years the browser quietly...
信息来源:DEV Community
教程 / 实战普通
Boris Cherny's Steps of AI Adoption, Through the Eyes of Our Pipeline
You run Claude Code, or something like it, on a real codebase. Your CLAUDE.md grew one line at a...
信息来源:DEV Community · Claude
教程 / 实战普通
Does your WebMCP actually save time and tokens? Run an ON/OFF experiment
I added a WebMCP benchmark to DeepDeck, my MIT-licensed macOS desktop project built on DeepSeek...
信息来源:DEV Community · DeepSeek
教程 / 实战普通
-preset slow bought 0.16%. Ten ffmpeg settings measured on two ARM cores.
I am Obole, an AI. I run on a two-core ARM server with no GPU, I measure the tools I actually use to...
信息来源:DEV Community
教程 / 实战普通
Machine-Speed Credential Abuse: What the ChainDrop npm Worm Changes
Machine-Speed Credential Abuse: What the ChainDrop npm Worm Changes ChainDrop hijacked 444...
信息来源:DEV Community
教程 / 实战普通
I almost started a Medium column about getting cited by LLMs. Then I read Medium's robots.txt
I was about to start a monthly column about AI search visibility — how to get your pages cited by...
信息来源:DEV Community
教程 / 实战普通
Tool Descriptions Are the Contract
This series started with a claim: MCP and REST are two doors into the same kitchen. Three posts...
信息来源:DEV Community
教程 / 实战普通
My MRI interpretation vision model pointed at the right kidney and wrote "left"
I build a tool that reads MRI and CT scans with general vision models and explains the findings in...
信息来源:DEV Community
教程 / 实战普通
Show a model your old code and it writes your old bugs: 32 runs, 0% reuse
Same task, same models, two commits of my own repo. From the pre-migration commit every run hand-rolled a 190-line typeahead with the exact defects the migration removed; from the post-migration commit every run reused the shared component…
信息来源:DEV Community
教程 / 实战普通
I built an LLM tuner. Benchmarking it proved me wrong four times.
My LLM tuner picked a configuration that ran almost four times faster. I had passed a flag to limit...
信息来源:DEV Community
教程 / 实战普通
Google Research 推出 Retrieve-for-Train(R4T):用 RL 编译的扩散检索器实现 12× 至 20× 更快的查询扇出
Google Research 提出 Retrieve-for-Train(R4T),先用强化学习离线学习查询扇出,再蒸馏进一个 53.9M 参数的扩散 Transformer,单次非自回归前向即可生成全部检索方向。
信息来源:MarkTechPost(RSS)
教程 / 实战普通
DeepSeek-V4.1-Flash 技术报告解读:CED 与 CSA2 如何将 KV 缓存压缩至每 token 890 字节
作者解读《DeepSeek-V4.1-Flash: Pushing the Limits of KV Cache Compression》技术报告,指出该模型虽名为 V4.1 Flash,但架构变化足以视作 V5 级迭代。
信息来源:Hacker News 热门(buzzing.cc 中文翻译) · DeepSeek
教程 / 实战普通
GPT-6 Astra 联手 Tripo 3.1 自动搭建 GTA 6 风格海滩别墅
用户 insaneUEFN 用 OpenAI GPT-6 Astra、Atlas3DAI 和 Tripo 3.1 搭建了一条自动化管线,将 GTA 6 风格海滩别墅直接生成并组装进 UEFN。他放弃低模的 Tripo P2 改用 Tripo 3.1,把 70 个高模资产优化到 5K 或 1K 三角面,再通过 UEFN MCP 复用梁柱、窗户等模块化部件完成 1:1 重建,全程仅需两次方向性反馈。
信息来源:Tripo(官方 X) · OpenAI / GPT
教程 / 实战普通
Freeze a Holdout Before You Quote a Coding-Agent Score
You are in a Friday ranking review. Someone pastes a table: Agent B hit 71 percent. Last week it was...
信息来源:DEV Community
教程 / 实战普通
用GPT6加Hyper3D Rodin搭建可互动3D个人作品集的实操教程
作者用GPT6搭配Hyper3D Rodin的MCP,花一下午做出了可逛、可交互的3D个人作品集网页,场景含35个精细建模。流程为GPT-6画参考图并写交互代码,Rodin生成3D模型,BANG把模型按结构拆成独立零件以实现零件级动画,作者还把整个流程打包成skill发布在github.com/LearnPrompt/3Dworkshop。
信息来源:公众号:卡尔的AI沃茨 · GPT
教程 / 实战普通
圆谷公司谈AI如何变革视频创作
AI 正在为视觉叙事开启新篇章。在 #ApsaraConference2026 上,Tsuburaya Fields Holdings 执行董事 Toyoshima Yusaku 将探讨视频创作如何随 AI 演进。立即注册:https://click.alibabacloud.com/m/20000002869/ #AgenticEra #AgentNative #AIAgentsAtApsara #BringYourAgent #AlibabaCloud
信息来源:X:阿里云 / Alibaba Cloud (@alibaba_cloud)
教程 / 实战普通
Muse 模型用于工业领域开发
用 muse 来构建工业级的东西! 【引用 @BalaSelvam】:正在为工业领域打造一些非常酷的东西,我们用 Muse 作为编排器,我只能说哇,太哇了。太惊艳了;Muse 系列模型没有让人失望,我每一分钟都很享受。
信息来源:X:Alexandr Wang(Scale AI 创始人/Meta 首席 AI 官) (@alexandr_wang)
教程 / 实战普通
The idle-parent trap on LLM Agents
Companion to Two agents, one night, 158 million tokens. About four minutes to read. The...
信息来源:DEV Community
教程 / 实战普通
We put an ONNX background-removal container behind a Next.js Worker on Cloudflare
We put an ONNX background-removal container behind a Next.js Worker on...
信息来源:DEV Community
教程 / 实战普通
I Let AI Plan 170 Changes. It Made the Same 3 Mistakes Every Time.
Everyone is arguing about which model plans best. I ran 170 goals and found out the model was never...
信息来源:DEV Community
教程 / 实战普通
TechCrunch Disrupt 2026:Gusto、Insight Partners 与 Leland 谈 AI 智能体如何改变初创团队招聘
TechCrunch Disrupt 2026 的 Builders Stage 将举办"Hiring When AI Is a Co-Founder"对谈,Gusto CEO Josh Reeves。
信息来源:TechCrunch:AI(RSS)
教程 / 实战普通
Two agents, one night, 158 million tokens: what it cost and what we are fixing
Draft for publication · 16 September 2026 · Owner: Aashish Bhandari (Max) · Author: Claude ("Naruto")...
信息来源:DEV Community · Claude
教程 / 实战普通
I wired a real fruit fly brain to reply to my emails (no cloud API)
Last week Google/Janelia released the complete wiring diagram of a fruit fly brain, and someone...
信息来源:DEV Community
教程 / 实战普通
My First AI Judge Interview: What Could Possibly Go Wrong?
TL;DR: I entered HackerRank Orchestrate wanting to win, then faced an AI judge asking how Praxi Clew...
信息来源:DEV Community