AI 编程

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：Discussion | Link

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：Checking agent-generated code usually means hopping between tabs. Learn how to view diffs, run terminal commands, and preview web apps side by side in the GitHub Copilot app. The post GitHub Copilot app for Beginners: Using the diff, terminal, and browser appeared first on The GitHub Blog.

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：Google has open-sourced Mantis, a stack-agnostic toolkit of security review skills for AI coding agents. It runs the full vulnerability lifecycle: sweep the code, filter false positives, reproduce the bug in a sandbox, patch it, re-attack the patch, then score the risk. Apache 2.0, and documented as demonstration-only. The post Google Open-Sources Mantis: A Modular Skills Toolkit That Lets Coding Agents Find, Reproduce and Patch Vulnerabilities appeared first on MarkTechPost.

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：AI models now power all manner of agents, from coding assistants that write and debug software to customer service systems The post Claude did best on a new benchmark for agents that build agents. It still passed fewer than a quarter of the tests. appeared first on The New Stack.

微软Edge团队承认，AI辅助编码让浏览器扩展的提交量激增，使得审查管道承受额外压力。为应对这一局面，微软将自动化重复性验证检查，并宣称不会降低审查标准和安全性，还计划每15天更新一次“精选”徽章。此前，微软Exchange团队也因AI发现过多bug而推迟了累积更新。

本文介绍五种无需付费订阅或 GPU 即可使用 AI 编程代理与模型的方法：OpenCode Zen 的免费模型、ChatGPT 免费计划中的 OpenAI Codex、Kilo Code 的 Auto Free、OpenRouter 的免费模型 API，以及 Google Antigravity 的 $0 计划。包含安装命令、使用要点与注意事项。

Catenary 是一款面向 AI 编程代理的空间画布 IDE，将代码与 AI 工作流可视化地组织在无限画布上。产品目前通过 Product Hunt 发布并设有讨论区。

本文提出了Harbor Adapters，一个统一的智能体基准评测基础设施，可将80多个基准适配到任意智能体上；通过代码审查与对比实验验证后，研究团队在54个基准上对8个模型进行了大规模评测，并配合Terminus-2与3种原生评测框架运行。进一步，团队推出了Harbor-Index，这是从适配基准中提炼出的精选元数据集，包含跨越29个基准的82个困难、多样且高质量的任务。即使最强配置（GPT-5.5 with Codex）的通过率也只有28.0%，能够以可负担的成本保留大规模智能体评测的挑战性与覆盖面。相关适配器、结果、分析和数据集均已开源。

Project HydraFusion 是 GitHub 发布的研究预览，它不再将模型选择视为一次性设置，而是针对每个请求构建并动态编排一个执行工作流，可在不同提供商的模型间进行起草、批判、升级等操作。目前仅在 GitHub Copilot CLI 中以研究预览形式提供，并按实际调用的各模型标准令牌费率计费。

Simon Willison 分享了一则 TIL：在 macOS 上，ChatGPT Codex 等编程智能体可以直接调用已安装的 Blender 应用来生成 .blend 文件、渲染静态图像与影片。他用“鹈鹕骑自行车”的场景演示了如何通过几句自然语言提示和后续迭代获得精致的三维插画。

Coder发布Coder Agent Relay服务，以SpaceXAI为首个合作伙伴，让受监管行业的企业可以自托管运行编码智能体，同时由Cursor在云端负责推理和规划。Coder CEO指出，真正的障碍不是token花销，而是技术采用不均衡和部署架构不匹配。该服务目前处于私有预览阶段。

GitWarren 是一款面向开发者的工具，让开发者在提交代码变更之前，通过自己的编码智能体对代码进行审查，以便提前发现并改进问题。

数据转换是将多个来源的原始数据转化为标准化、高质量数据集的过程，使其适合存储、分析和决策。本文介绍数据转换的重要性、与ETL的关系、常见转换类型、性能优化方法，以及批处理/流处理、本地/云、开源/商业、无代码/低代码等工具选型维度。

了解如何在 GitHub Copilot 应用中并行运行多个代理，体验这种从感到混乱到掌控全局的转变。每个代理会话独立运行，互不干扰，让你在更短时间内完成更多任务。

这篇论文提出信念校准优化（BCO），让编码式优化代理将隐性的环境信念写为持续更新的上下文文档，作为可复用的显式世界模型。在记忆问答、工具使用问答、代码型应用代理和终端代理等五组基准上，BCO较仅缺少该模块的对照方法获得更高训练通过率；该优势在未参与候选选择的保留集上依然成立。后续换用不同基础模型及离线内容消融进一步说明，收益源自文档携带的信息内容。

Gemini 3.8 Flash 被定位为迄今最智能的实用型模型，专注于编程与智能体场景，兼具日常部署的可靠性与更强的代码和工具调用能力。

Sidebranch 是一款简单的可视化差异对比工具，基于 Git 工作流，帮助开发者更直观地查看代码分支与提交之间的变化。

本文介绍如何使用 Amazon Bedrock AgentCore 构建自动化架构文档流水线。全球银行间经纪商通过 AgentCore、Amazon Bedrock Knowledge Bases、AWS CodePipeline 等 AWS 服务，自动分析 .NET 代码库、生成并校验 Mermaid/UML 架构图、存入 S3 并支持语义检索，实现持续更新的可搜索架构文档，生产可靠性约 95%，自 2026 年第一季度投入运行。

GitHub 博客介绍了 Copilot 降低 AI 编码成本的四个实践：不要只看单次工具调用的 token 数，而应从完整任务出发。通过选择性压缩噪音输出、删除不再使用的行号格式、用行为测试保护提示词压缩、以及批量直接交付后台完成结果，团队在离线/在线实验中让推理成本或 token 用量下降约 1–5%，且未出现质量回退。

Paint.NET 作者 Rick Brewster 透露，为了让 Paint.NET 在 WINE 上运行，他们使用 AI 工具 Claude 从零开始逆向工程重写了 Direct2D。这个 18 万行的代码库被戏称为“vibe coding”产物，未经全面审查，但 Claude 的逆向工程能力令人印象深刻，同时也需要大量人工监督。

Anthropic 推出 Claude Fable 5.1，内置文本水印，但该水印在代码等低熵场景中效果有限，且新 API 账户的思维块保留受限，给开发者带来新的权衡。

Meta正式将Muse Code从测试版中推出，距首次亮相不到一个月。新增功能包括会话间消息传递、工作流和倒带功能，并推出了三个订阅计划：$5、$15和$50/月。价格远低于竞争对手，但早期测试显示其代币消耗较高且代码重构结果较弱。一些专家担心Meta的商业模式可能利用数据来投放广告。

shaide 是一个自托管的 AI 平台，可部署在您自己的 Kubernetes 集群上，实现大规模多模型推理。它通过单一命令安装，完全在您的网络边界内运行，支持无互联网连接的环境。该平台以基础设施即代码（Pulumi）管理，提供 OpenAI 兼容 API，专为代理集群设计，并适用于多家云提供商和本地环境。

Foremerge是一个开源协调协议，通过让AI编码代理在编辑前声明意图，防止它们之间的冲突。它构建在Git之上，使用共享数据库对计划进行语义比较，在代理编码前发出冲突警告，不锁定文件也不依赖模型判断。

AI工作流中的错误往往是静默的：训练代码顺利运行并输出合理数值，但结果却不可信。本文总结了七个常见的Python陷阱，包括预处理泄漏、错误的随机分割、训练与推理代码不一致、种子设置不完整、混淆eval和no_grad、广播导致形状错误，以及将模型文件视为惰性数据。每个问题均附有检测方法和实践建议。

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：Back to Blog The Problem Modern AI work already spans multiple tools. You might research in Gemini, think things through in ChatGPT, build in Claude Code or Cursor, and keep follow-up discussions going across all of the…

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译： Introducing wrapture New from Graham Dumpleton (of wrapt, mod_wsgi, and New Relic's Python agent fame), who describes Wrapture as taking the monkeypatching ideas from wrapt and extending them to apply to testing and tracing at the same time. Wrapture (full documentation here) makes it easy to wrap any function or method such that all access can be traced, or can be overridden to return a different value. It acts as both an alternative to unittest.mock and a way to implement tracing against an existing project: Attaching observation to code you do not control, recording what flows through it, and doing so without disturbing the program being watched, is a problem I have never really stopped thinking about. Wrapture includes OpenTelemetry support and even has an…

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：Three weeks ago we published an audit tracing a widely repeated 97%-failure-rate statistic back to a content-marketing article with no source, and along the way found a data-integrity bug in the most-reused public n8n w…

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：Cursor began rolling out Origin, its own code hosting platform, to paid users on Monday morning. Roughly three and a half hours later, GitHub's status page lit up with what became a six-hour-and-forty-two-minute global…

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：hotSecurity Aug 30, 2026 • 3 min read AI coding agents followed abandoned package references Researchers found 120 unclaimed package or domain references across corporate llms.txt files. AI agents could use them to inst…

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：Tags: AI, LLMs, Coding, Technical Debt, Software Engineering Superhuman AI coding still makes software worse. AI is probably better than humans at writing code now. But skill at coding is not exactly the same thing as s…

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：NatashaTheRobot Aug 30, 2026 I’ve been seeing a lot of posts lately saying that software engineers who use AI miss the flow state. This is a bit surprising to me because if anything, using AI puts me into an ULTRA INTEN…

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：MCPSTRICT TYPESCRIPTZERO-CONFIG Persistent, Project-Local Memory for AI Coding Agents Coding agents forget decisions between sessions. OpenContext MCP exposes a lightweight Model Context Protocol server that enables AI…

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：AI-assisted development Use Cursor, Claude, Copilot, or any coding agent to build on Lumify — with MCP tools, machine-readable docs, and copy-paste prompts that prevent hallucinated endpoints. API key Sign in, or get an…

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：Elon v Altman has a real consequence.

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：An AI copilot is an AI assistant embedded directly in a software application. It...

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：Uh oh! There was an error while loading. Please reload this page. Notifications You must be signed in to change notification settings Fork 0 Star 1 BranchesTags Open more actions menu Latest commit History 29 Commits 29…

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：AI coding bot vendors tell you to use a context file with instructions for the chatbot on how to edit your project. Claude Code wants a CLAUDE.md, or there’s AGENTS.md in general. [Anthropic] But does your AGENTS.md do…

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：Notifications You must be signed in to change notification settings Fork 0 Star 1 BranchesTags Open more actions menu Latest commit History 8 Commits 8 Commits Folders and files NameName Last commit message Last commit…

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：Notifications You must be signed in to change notification settings Fork 1 Star 8 BranchesTags Open more actions menu Latest commit History 79 Commits 79 Commits Folders and files NameName Last commit message Last commi…

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：Notifications You must be signed in to change notification settings Fork 0 Star 1 BranchesTags Open more actions menu Latest commit History 177 Commits 177 Commits Folders and files NameName Last commit message Last com…

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：The product is the crane A long-horizon harness you run. Not a plugin pack inside someone else’s. Most things branded “harness engineering” are skills, agents, and slash commands that sit inside Claude Code or Copilot.…

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：- YouTube AboutPressCopyrightContact usCreatorsAdvertiseDevelopersTermsPrivacyPolicy & SafetyHow YouTube worksTest new features

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：AI coding company Replit is throwing its weight behind the model-routing trend by making its “intelligent model routing” system the The post Replit’s new default: Auto mode picks the best model for each task appeared first on The New Stack.

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：The Independent AI Coding Community AI Tools Search & browse all AI tools AI Jobs International roles · opportunities Creative Studio Image · Video · Training AI Models Curated models, explained AI Skills Handy prompts,…

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：Managing library updates can be tedious at times. Learn how the GitHub Copilot app can handle this type of repetitive task. The post GitHub Copilot app for Beginners: Automate Dependabot pull request triage appeared first on The GitHub Blog.

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：Everyone's using AI coding agents. Here's how to make yours actually useful.

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：Uh oh! There was an error while loading. Please reload this page. Notifications You must be signed in to change notification settings Fork 0 Star 1 BranchesTags Open more actions menu Latest commit History 8 Commits 8 C…

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：arXiv:2608.23622v1 Announce Type: new Abstract: Large language models (LLMs) have shown strong capabilities in reasoning, planning, and tool use, but many scientific and engineering tasks require more than plausible text and code generation. They require understanding how a system responds to intervention, which in practice depends on controlled experimentation. In this work, we propose a multi-agent framework that enables LLM agents to conduct controlled experiments with scientific simulation models for pharmaceutical process design. Given a user query and a baseline configuration, the system constructs a structured task representation, designs experiments, executes comparative simulation, interprets the resulting outcomes, and synthesizes evidence-based recom…

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：TL;DR — Key Takeaways JetBrains has launched Junie Local, an on-device version of its AI coding agent that runs entirely on supported Macs without cloud inference or token charges. Junie Local uses a JetBrains-tuned, 4-…