多篇文章讨论了 Claude Code(一款 AI 编码助手)的增强功能和最佳实践。一篇文章介绍了 `az-brake`,这是 Claude Code 的一个插件,为任何会产生费用的 Azure CLI 命令添加了一个关键的预命令审批步骤,旨在防止意外计费。另一篇文章介绍了 `claude-mem`,一个为 Claude Code 提供会话内存的开源工具,使其能够回忆过去的决策和上下文。其他文章深入探讨了如何管理 Claude Code 的决策,例如通过更新或删除过时的文件来防止其遵循过时的指令,以及在 2.1.323 版本中引入的实验性“代码模式”,该模式允许 Claude 编写 JavaScript 来在单个步骤中执行多个工具调用。`CLAUDE.md` 作为 Claude Code 项目指南的重要性也得到了强调,它充当了可以通过 `/init` 或 `/plan` 命令启动的入职文档。最后,解释了子代理的概念,其中具有自己上下文窗口的辅助代理可以执行辅助任务并将摘要返回给主对话,从而可能优化令牌使用。
AI
影响
这些发展旨在提高 AI 编码助手的安全性、内存和效率,从而可能加速开发人员的工作流程并减少错误。
<p>Map the codebase once, then ask questions against the map. Have AI write down what the code does, its modules, the riskiest files and the rules you mustn't break, in one file in the repo. Every answer after that comes with the file and line, so you can check it. The <a href="h…
dev.to — Claude Code tag
TIER_1English(EN)·Eventine·
<p>For the second week in a row I had watched my Claude Code usage climb, and I had no idea why. Then I answered one of my sessions, got up for a drink, and came back a few minutes later to find my 5-hour usage far higher than I had left it.</p> <p>That was when I started looking…
dev.to — Claude Code tag
TIER_1English(EN)·Dmitriy Solodukha·
<p><em>Claude Code GUI options: Anthropic's desktop app, VS Code and web UI; open-source opcode, CloudCLI and Nimbalyst; and Caprock, a free app for every agent.</em></p> <p>Yes. Anthropic's official GUI for Claude Code is the <strong>Code</strong> tab of the<br /> Claude desktop…
dev.to — Claude Code tag
TIER_1English(EN)·laogedev2·
<p>A Claude Code skill is mostly a Markdown file — but once installed, it runs with your tools, reads your code, and can tell the agent to execute things. I install skills from community collections all the time, and before any of them lands in <code>~/.claude/skills</code>, I ru…
dev.to — Claude Code tag
TIER_1English(EN)·mustafa morbel·
<blockquote> <p><em>Originally published on Medium (JavaScript in Plain English). <a href="https://javascript.plainenglish.io/taming-claude-code-a-guide-to-claude-md-and-hooks-ed059879991c?source=friends_link&sk=aed227294179c446ff1d1f288ac135d6" rel="noopener noreferrer">Read…
dev.to — Claude Code tag
TIER_1English(EN)·Muhammad Mairaj·
<p>An AI agent can create a billable Azure resource in one line. The only brake most setups have is you, reading every command before it runs, every time, forever.</p> <p>I stopped trusting that brake the day it worked exactly as designed.</p> <p><strong>The approval was a conver…
<h2> What it is </h2> <p>claude-mem is a free, open-source tool that gives your coding agent a memory across sessions. A coding agent is a program that writes and edits code for you from plain prompts. Each new session starts blank. The agent re-reads the same files, re-asks the …
dev.to — Claude Code tag
TIER_1English(EN)·Alfredo Izquierdo·
<p><em>Originally published on <a href="https://contextforge.dev/blog/stop-claude-code-from-following-outdated-decisions" rel="noopener noreferrer">contextforge.dev</a>.</em></p> <p>I spent this week trying to stop Claude Code from following outdated decisions, and the first thin…
<p>Claude Code 2.1.323, released on 12 November 2026, adds an experimental, opt-in Code Mode in which Claude writes JavaScript to call its tools, and it ships alongside dozens of Bash permission and sandbox fixes. Our verdict: standard tool calls win for day-to-day work today, an…
HN — claude cli stories
TIER_1English(EN)·craigmart·
<p><strong>In short:</strong></p> <ul> <li>"What should I do first?" In the three projects I started from an empty directory, the first step was always CLAUDE.md.</li> <li>I created the first version in three different ways: writing the requirements myself over about four hours, …
dev.to — Claude Code tag
TIER_1Français(FR)·System Design Simple·
<p>Claude Code subagents let one agent hand reading work to helpers that each run in their own context window and send back a short summary, so the main conversation keeps the task instead of every file it read. Each helper adds requests and overhead; total token use depends on h…
dev.to — Claude Code tag
TIER_1English(EN)·Joan Comadran·
<p><strong>Code Mode is an experimental Claude Code switch, off by default, where Claude writes JavaScript that fires several tool calls in one step and returns only the output it needs.</strong> Turn it on with <code>CLAUDE_CODE_EXPERIMENTAL_CODE_MODE=1</code> in a terminal sess…
dev.to — Claude Code tag
TIER_1English(EN)·Lookbk·
<p>You start a refactor at your desk, then have to leave. The tests are still running and Claude is waiting on a decision. You want to check in from your phone without moving the work to someone else's server. That is the problem Claude Code Remote Control solves. It connects cla…
dev.to — Claude Code tag
TIER_1English(EN)·Tarun Agarwal·
<p><em>Six small files that turn a coding agent from a clever stranger into a teammate who knows how your project works.</em></p> <p>Day one with Claude Code goes like this.</p> <p>You explain your project. Tests live in <code>/tests</code>, not next to the code. We use <code>pnp…
dev.to — Claude Code tag
TIER_1English(EN)·Tanishq Agarwal·
<p>After building dozens of custom skills and subagents in Claude Code, two problems kept biting me:</p> <ol> <li>Always-loaded skill definitions were burning 7,300+ tokens before I even typed a prompt.</li> <li>A skill that worked great on Opus would quietly degrade on Sonnet or…
dev.to — Claude Code tag
TIER_1English(EN)·Roi Dayan·
<p>"You've hit your session limit" is the line nobody wants mid-task. Here's how Claude Code's limits actually work on a Claude plan, how to see one coming, what your options are when you hit it, and how to make your usage last longer. Every fact is from Anthropic's own pages, ch…
dev.to — Claude Code tag
TIER_1Português(PT)·Tiago Vilas Boas (Montanha)·
<p>Quando estava construindo o <a href="https://github.com/tiagovilasboas/downshift-ai" rel="noopener noreferrer">Downshift</a>, precisei entender melhor como o Claude Code gastava contexto e escolhia os modelos dos subagentes. Eu vinha de uma dor bem concreta: pagava o plano Pro…
dev.to — Claude Code tag
TIER_1English(EN)·Eren Çolak·
<p>I usually have four or five Claude Code sessions open at once, and my problem was never the agents. It was me: a session would sit on a permission prompt for twenty minutes in a tab I wasn't looking at, or two sessions would edit the same file and one would quietly undo the ot…
dev.to — Claude Code tag
TIER_1English(EN)·Naman Jain·
<p><em>Originally published at <a href="https://www.getlockstep.dev/blog/best-memory-tools-claude-code" rel="noopener noreferrer">getlockstep.dev</a>.</em></p> <p>Claude Code already has memory. Before installing another tool, work out what it is failing to remember, who needs th…
dev.to — Claude Code tag
TIER_1English(EN)·Fred 🍓·
<p>Most Claude Code tooling I read about assumes macOS or Linux. I build a Windows desktop app that starts Claude Code sessions in terminals, writes commit messages with <code>claude -p</code> and runs hooks all day. Here's what went wrong on Windows, in the order I hit it, so yo…
dev.to — Claude Code tag
TIER_1English(EN)·Roi Dayan·
<p>If you've never written code, and the idea of starting makes you a little nervous, this guide is for you. You don't need to know anything about programming, and you don't need to be "a computer person". If you can write a text message, you can do this.</p> <p><strong>Vibe codi…
dev.to — Claude Code tag
TIER_1English(EN)·Arooba Fatima·
<p>When you jump into a codebase you've never seen before, you spend the first few hours just trying to figure out what everything is. Which file does what. How the pieces connect. Where the data actually flows.</p> <p>I got tired of that, so I built <strong>Floorplan</strong> — …
dev.to — Claude Code tag
TIER_1English(EN)·Naman Jain·
<p><em>Originally published at <a href="https://www.getlockstep.dev/blog/claude-code-memory" rel="noopener noreferrer">getlockstep.dev</a>.</em></p> <p>On Monday you tell Claude Code to use pnpm, not npm. On Tuesday it uses pnpm without being asked. On Wednesday your teammate ope…
dev.to — Claude Code tag
TIER_1English(EN)·SoluTerminal·
<p><em>First posted on X. Updated for the current version.</em></p> <p>I run Claude Code and Codex side by side on Windows all day, mostly inside WSL. Most of what goes wrong isn't the<br /> models. It's the plumbing around them. Here's what breaks, and what I do about it.</p> <h…
dev.to — Claude Code tag
TIER_1English(EN)·Lightning Developer·
<p>Access to a real terminal from any browser is one of those superpowers that sounds too good to be true. You keep your workstation or server where it is, you leave a small agent running, and then from a laptop, tablet, or even a borrowed computer you open a browser tab and get …
dev.to — Claude Code tag
TIER_1Español(ES)·Nixon Perinchery·
<p><strong>The short version</strong></p> <ul> <li>In a large repository, keep Claude's context on the part of the code a task touches: unrelated instructions and file reads cost tokens and can degrade Claude's performance.</li> <li>Where you launch claude decides which files it …
dev.to — Claude Code tag
TIER_1Nederlands(NL)·Amar·
<p><em>Cross-posted from <a href="https://iqraa.tech/ai-genai/claude/claude-code-subagents/?utm_source=devto&utm_medium=referral&utm_campaign=claude-code" rel="noopener noreferrer">Iqraa.tech</a> — hands-on tutorials for Claude Code, AI agents, cloud, and system design.</…
<p><em>Cross-posted from <a href="https://iqraa.tech/ai-genai/claude/claude-code-extensibility/?utm_source=devto&utm_medium=referral&utm_campaign=claude-code" rel="noopener noreferrer">Iqraa.tech</a> — hands-on tutorials for Claude Code, AI agents, cloud, and system desig…
Medium — Claude tag
TIER_1English(EN)·Millroy Fernandes·
<div class="medium-feed-item"><p class="medium-feed-snippet">The built-in commands, CLAUDE.md changes and model settings that make your plan last until the reset, all checked against Anthropic’s own…</p><p class="medium-feed-link"><a href="https://medium.com/@grokyb…
<p>Claude Code reaches a usage limit while you are debugging. Your next assistant needs the failing command, the approaches you rejected, and the project directory where the work happened.</p> <p>This walkthrough uses an illustrative debugging session. I maintain <a href="https:/…
Medium — Claude tag
TIER_1English(EN)·Bobo Zivkovic·
<div class="medium-feed-item"><p class="medium-feed-snippet">Why your skill never triggers, how to call one by name, and the one field in SKILL.md that decides everything</p><p class="medium-feed-link"><a href="https://bobozivkovic.medium.com/how-to-use-agent-skills-in-claude-cod…
<h4>A year of building across five projects, losing track of sessions, sharing my desktop with AI, and learning where the time and tokens went.</h4><figure><img alt="Editorial illustration of five different software projects and AI coding windows surrounding a laptop." src="https…
Medium — Claude tag
TIER_1English(EN)·Smallbizsense Ai·
<div class="medium-feed-item"><p class="medium-feed-snippet">Claude Desktop offers one of the cleanest AI workspaces available, complete with Projects, Artifacts, and a strong coding harness until…</p><p class="medium-feed-link"><a href="https://medium.com/@smallbizsense.a…
<p>Ask Claude to calculate the standard deviation of a column in a spreadsheet, and it doesn’t reason through the arithmetic in plain text. It writes a few lines of Python, runs them, and reads back the result. That write-run-read loop is the entire premise of Anthropic’s code ex…
dev.to — MCP tag
TIER_1English(EN)·Dinesh Rajendran·
<p><strong>Agent memory works best when it's on a diet: inject only what the agent can't find out for itself, and let it ask for the rest.</strong></p> <p><strong>Key findings</strong></p> <ul> <li>Inject only what a coding agent can't find out for itself: decisions and rejected …
Medium — Claude tag
TIER_1Nederlands(NL)·techpotions·
<div class="medium-feed-item"><p class="medium-feed-snippet">/sandbox isolates Claude Code’s Bash commands at the OS level. Writes stay in your project, network traffic goes through an allowlist, and…</p><p class="medium-feed-link"><a href="https://medium.com/@janja…
<div class="medium-feed-item"><p class="medium-feed-snippet">The Reddit threads asking “anyone using Claude Code for non-coding tasks?” are long because the answer is yes, with caveats nobody puts in…</p><p class="medium-feed-link"><a href="https://medium.com…
<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*E04RwYxPRLJpbhHvWuyHvw.png" /></figure><blockquote><strong><em>Key Takeaways</em></strong></blockquote><blockquote>- OpenSpec is a free command-line tool that makes your coding agent write a short plan before it …
<p><em>This is an English version of an article first published in Japanese on <a href="https://zenn.dev/murooka_az/articles/failure-scan-two-claude-code-bugs" rel="noopener noreferrer">Zenn</a>.</em></p> <p>When the same failure keeps happening in Claude Code, it's tempting to a…
<p>Claude Desktop offers one of the cleanest AI workspaces available, complete with Projects, Artifacts, and a strong coding harness until you hit a message cap mid-build. The free plan runs out within minutes, the $20 Pro tier can disappear in a single afternoon, and even the $2…
<p><em>Originally published on the <a href="https://quicksilverpro.io/blog/claude-code-quicksilver-pro/?utm_source=devto&utm_medium=referral&utm_campaign=claude-code-setup" rel="noopener noreferrer">QuickSilver Pro blog</a>.</em></p> <p><a href="https://code.claude.com/" …
<p><em>Originally published on the <a href="https://quicksilverpro.io/blog/claude-code-github-codespaces-vs-machines/?utm_source=devto&utm_medium=referral&utm_campaign=codespaces-vs-machines" rel="noopener noreferrer">QuickSilver Pro blog</a>.</em></p> <p>GitHub Codespace…
A user-level Claude Code config: pinned settings, a required sandbox, lean instructions. - Fewer tokens per session - Secrets out of Claude's reach - Same behaviour on every machine Nothing project-specific loads by default. Pinned means explicit settings and a minimum version on…
I find that using #claudecode via a remote control server is the _best_ way to use #claude . I have a powerful workstation with claude where it has excellent bandwidth to the cloud. Then I control the #ai from my phone or laptop even in low-bandwidth settings. blog.lhotka.net/202…
<!-- SC_OFF --><div class="md"><p>Hey guys, I have been blindly using Cursor for the past year and a half, so I haven't understood the premises of any of the existing alternatives.</p> <p>My existing preconception is that Claude Code is a terminal tool with no IDE to look at the …
<!-- SC_OFF --><div class="md"><p>I run a small agency and use Claude Code for pretty much everything.</p> <p>That includes client projects with their own deployment pipelines and GitHub versioning across several servers, plus all the daily agency work like emails and to-dos.</p>…
<!-- SC_OFF --><div class="md"><p>I've been using both the regular Claude chat and Claude Code for a while, and I keep running into the same pattern. In the chat, I end up bouncing between conversations: a thread gets too long or drifts, I open a new one, re-explain the context, …
<!-- SC_OFF --><div class="md"><p>As a developer, I intentionally maintain separate configurations for Claude's web chat, which I use primarily for personal purposes, and my local CLI development environment.</p> <p>Today, I discovered that Claude Code had automatically synced ne…