Community posts

为什么不整个传奇网游mod,等claude结果的时候进去一刀9999,不比这解压

Claramente de las mejores features los mods de Claude code. Agarre cositas de la gentle shell y orca para trackear mejor las sesiones. Colores customizables para la pipol. Le daré un par de vueltas y la largaré a la comunidad.

フォークしたら、dmca冤罪にされちゃいそうなのでできない https://claude.com/blog/claude-code-mods

【小学生でもわかるAIニュース速報】No.054 AIの仕事場を、自分の仕様に。 「Claude Mods」登場！ ✅ 見たい情報を、会話の横に表示 ✅ 作業の途中に、確認や処理を追加 ✅ 欲しい拡張をAIと相談して作り、共有 /diffやAGENTS.md対応もModsで作られています。 私たちの仕事がどう変わる？ 使い道・料金・始め方を図解でやさしく解説しました。 記事はこちら👇 https://x.com/engineer_num1/status/2106351136873808252 #ClaudeCode #ClaudeMods #AIニュース

Claude Code est maintenant moddable 🧩 Une phrase → Claude écrit le mod, et il se recharge à chaud. J'en ai fait 2 en live : 🛡️ un garde du corps (adieu les rm -rf) 🚀 un cockpit live dans l'interface La vidéo 👇 https://youtu.be/_Hllg5sDhFo

Funny how people are going crazy over Claude Mods when terminal users have had it for months now 😂

Made three mods for Claude Code: plan-bar: progress bars while Claude works through a plan leftovers: tracks containers and services Claude left running, so you can clean them up secret-mask: masks tokens in command output before Claude reads them https://github.com/homieyangg/claude-code-mods

Claude Code kept retrying the same failing fix, so I built a mod that catches it 🔁 It flags the loop, explains why, and Rethink makes Claude change approach. Runs locally. No extra model calls. (Loop staged for the demo.) https://github.com/ktripathi2281/LoopDetector

有人在终端里实现了带声音的视频播放：一个 Claude Code 模组把任意视频逐帧渲染成 ASCII 艺术。每帧包含 19.5 万个“像素”，每个字符同时带前景色和背景色，用来提升画面还原度。这更像是编码代理可扩展性的展示，实际清晰度和流畅度尚待验证。

Claude Codeに「mods」が登場。TypeScriptの小さな関数で、プロンプト書き換えやUI追加、コマンドのブロック、機密情報のマスクなど挙動を自由にカスタマイズ可能。プラグインで配布でき、企業向け管理機能も。 #Claude #ClaudeCode https://claude.com/blog/claude-code-mods

Claude Code mods make the agent runtime programmable: intercept tool calls, rewrite prompts, add UI. That also makes plugin review a security decision. Mods run unsandboxed with your machine permissions. https://claude.com/blog/claude-code-mods

Claude Code 的 mod，我还没装。先把官方写明的几句话摆出来。 mod 挂在 Claude Code 的每个事件上。 调用工具、请求权限、画界面，它都能看到。 官方博客写明，它能批准或拒绝权限请求。 权限和 Claude Code 一样，能访问你的整台机器。 换句话说，权限系统里多了一个能点允许的角色，而且是第三方写的。 Team 和 Enterprise 默认先加载防护 sec-default，拦 mod 覆盖你的拒绝规则。 没托管设置的机器没有这层，这是我按博客原文推出来的，博客没直接写。 所以没有托管设置的话，装之前那一步最要紧。 官方教程给了个办法，claude plugin validate。它不运行 mod，只列出它挂了哪些事件、调用了什么。 装之前先跑一遍，官方也说了，只从信任的来源装。 连官方示例都很坦白，Blast Radius 能在 rm -rf、git reset --hard 跑之前拦一下，教程却写明它只是个安全网。$(…)、别名和脚本里的 rm 都能绕过去，硬拦得靠权限规则。 我没装过 mods，没实测。 但我之前测 Coding Agent 何时停手，先发现的是评测环境没隔离干净。 环境不干净，后面的结果都不算数。 mod 就是你工作环境的一部分，装进去的东西不干净，之后 Claude Code 给你的每个结果，你都得多问一句能不能信。

Everyone is building Claude Code mods this week. Some are guardrails: block the push to main, block edits to .env. I built one on Claude Code's hooks. The spec I followed had the response format wrong. The gate would have said success and blocked nothing. Test the block.

Claude Code only shows "[Image #1]" when you paste an image. So I built a mod that shows a live thumbnail right in your terminal. What it does: 🖼️ Thumbnail above the prompt, with its number and size 🔍 /image-preview 2 opens a larger view

Artificial Intelligence Daily News · Saturday, October 3, 2026 1. Meta lets you build your own AI gadget Meta open-sourced firmware and a Linux SDK for Muse-powered gadgets on Raspberry Pi and ESP32 boards, plus 5,000 free Home Link dongles. No word on selling them. Sources: TechCrunch - https://techcrunch.com/2026/10/02/meta-wants-you-to-build-your-own-muse-gadget/ Engadget - https://www.engadget.com/2276312/meta-muse-gadgets-open-source-smart-home-link/ 2. Apple tightens what AI agents can read Apple will add controls so granting Full Disk Access, which exposes files, mail, messages and history, takes very explicit user action. No version or date given. Sources: Apple - https://developer.apple.com/news/?id=p6zjojqw TechCrunch - https://techcrunch.com/2026/10/02/apple-says-its-tightening-macos-full-disk-access-controls-due-to-new-risks-from-ai-agents/ 3. Suno reads your text aloud over music Suno’s Speech beta narrates a script over matching music, for poems, meditations and bedtime stories. Accents still slip, and training data is undisclosed. Sources: Suno - https://suno.com/blog/introducing-speech-beta The Decoder - https://the-decoder.com/ai-music-generator-suno-can-now-create-spoken-audio-with-matching-background-music/ 4. Anthropic lets coders rewrite Claude Code Mods are TypeScript functions that intercept Claude Code’s tool calls, permissions and rendering, so developers can add panels or swap commands. They run unsandboxed. Sources: Anthropic - https://claude.com/blog/claude-code-mods The Decoder - https://the-decoder.com/claude-codes-new-mods-system-lets-developers-rewrite-the-ai-coding-tool-from-the-inside/ 5. Anthropic will train 10,000 AI engineers Claude Frontier Academy commits $100 million to train 10,000 enterprise engineers by end of 2027, starting in San Francisco, New York and London. Entry is by nomination only. Sources: Anthropic - https://www.anthropic.com/news/claude-frontier-academy http://Unite.AI - https://www.unite.ai/new-anthropic-academy-backs-10-000-engineer-residencies-with-100m/ 6. Cloudflare opens two decision-making models Clef and Clef-flash score fixed options instead of writing text, with median results in 209 and 39 milliseconds, free under Apache-2.0. Figures are Cloudflare’s own. Sources: Cloudflare - https://blog.cloudflare.com/clef-decision-models/ The Decoder - https://the-decoder.com/cloudflare-says-its-new-clef-model-means-humans-no-longer-need-to-be-in-the-loop-for-ai-agents/ 7. GitLab patches a flaw in its AI gateway CVE-2026-90970 let authenticated Duo Agent users escape the prompt sandbox and run commands. Self-hosted AI Gateway needs 19.2.4, 19.3.2 or 19.4.1. None reported exploited. Sources: GitLab - https://docs.gitlab.com/releases/patches/other-patches/patch-release-gitlab-ai-gateway-19-4-1-released/ BleepingComputer - https://www.bleepingcomputer.com/news/security/gitlab-warns-of-critical-rce-vulnerability-in-ai-gateway-service/ #AI #LLM #AINews #Meta #Apple #Anthropic

Claude Code added mods on October 1: small TypeScript functions that rewrite a prompt, replace a built-in feature, or draw new UI, shipped inside plugins. Two details matter more than the launch itself. The built-in /diff is now a mod, and Anthropic says it plans to move more built-ins to mods so users can pare Claude Code down to a small core. The catch, from the same post: mods are not sandboxed and run with the same machine access as Claude Code itself, which is why a sec-default mod loads first on managed plans to block risky ones. https://claude.com/blog/claude-code-mods

The session-handoff plugin allows interoperability between Claude Code and Codex. The implementation is public at https://github.com/yuzushi-dev/session-handoff for anyone wanting to adapt it to their AI pipeline. #opensource

Claude CodeのgitクライアントMod実装お試し。 リアルタイムで機能が増えていく。 数時間で主要な機能はあらかた実装オワタ(´･_･`)

reminder this guy was using his early access as a creator to test his claude AI noshake and got mods/configs fried as a whole

@trq212 yeah mods is a great way to not manually manage my script now. now i am pushing all i have as plugins from and skill as i have had enough keeping it under private repo. https://x.com/VedantAndhale/status/2106316361052098733?s=20

Claude CodeでIMEなしに日本語入力でいるmodを作ってみた 慣れると快適そう https://github.com/skanehira/claude-vime

All claude code mods is here enjoy it https://claude.ai/artifact/6Tbij7Nmrz6Pa235tFvwfW

Claude Code just became a plugin host. Mods can rewrite prompts, block tools, and draw UI. They are not sandboxed. Same machine access as the agent. Install only what you trust. What would you mod first? https://x.com/i/article/2106220739170828288

Anthropic 给 Claude Code 上线了 Mods 系统，基于插件，开发者可以用 JavaScript 或 TypeScript 函数挂钩工具调用、用户提示和 UI 渲染等事件，也能加自定义面板、拦截工具调用或新增命令。 这类改动意味着 Claude Code 从「一个写代码的工具」往「可被改写的平台」挪了一步。以前想改它的行为只能等官方，现在钩子直接开在工具调用和 UI 层，插件生态的门就开了。对天天泡在里面的开发者来说，顺手程度会不太一样。 https://the-decoder.com/claude-codes-new-mods-system-lets-developers-rewrite-the-ai-coding-tool-from-the-inside/ #AI

記事書きました！ Claude Codeのmodsとはなにか - Think Twice https://thinktwice.tech/posts/df35487aab78004ebbdc05045f78590d/

今日のリポジトリ｜10/3（土） DietrichGebert/ponytail（★152,291） AIに「書きすぎ」をやめさせる、コーディングエージェント用のプラグイン できること ・コードを書く前に7段のはしごを順に当てる（そもそも要るか→既にこの中にあるか→標準の機能で済むか→入れてある部品で済むか→一行で済むか→最後に最小限） ・ブラウザやOSが元から持っている機能で足りる場面で、新しい部品を増やさせない ・検証・エラー処理・セキュリティ・アクセシビリティは削らせない（短くするために安全側を落とさない方針をREADMEで明記） READMEの計測では、Claude Code で実在のリポジトリを12件のタスクで編集させたとき、同じ条件の素のエージェントに比べてコード量が54%減・費用が20%減。既に最小限のコードではほぼ差が出ない、とも書かれています。 こんな人に：一言頼んだだけなのにファイルと依存がどっさり増えて、後から自分では何も直せなくなる人。 入れ方：Claude Code に「DietrichGebert/ponytail を入れて、書きすぎを抑えられるようにして」と頼めば進められます。 公式の手順は Claude Code で /plugin marketplace add DietrichGebert/ponytail と /plugin install ponytail@ponytail を順に送るだけ（Codex・Copilot CLI にも同じ形の手順あり）。 リンクはリプ欄に。

今週のClaude、新しいモデルが出たよ！ Sonnet 5.5、もう使えるよ。Anthropicによると、前のSonnetより30%以上速く答えるんだって。 Claude Codeにはmodsが来た！小さなプログラムを足して、動きや画面を自分好みに変えられるよ。 今日の動画はこちら📺 https://youtube.com/shorts/Qyig3wcJpkY #Claude #AI初心者

🔴 Right now on stream: Claude Code on Opus builds a mod for Quake × Counter-Strike 1.6. Build 17: "Counter-Strike 1.6 Takeover". ⏱ 55 min into this build. Next mix vote is open: burn $SIGF to pick the two games. It closes when the AI finishes this build. Watch live: http://sigf.ai

Built usage-bar, a tiny Claude Code mod 📊 One line above your prompt: • context used/total • 5h limit % + reset • weekly limit % + reset Live, configurable, zero tokens. /plugin marketplace add MohammadMD1383/claude-plugins /plugin install usage-bar@mohammadmd-plugins

@jarrodwatts Very nice! A few months ago I've done some surgery on various pieces of doom code and made it possible to run it as a tab next to the terminal. The server supports the real doom clients and web clients! https://github.com/fiorix/chan-ext-doom