Community posts

【海外で話題】Claude Codeの挙動やUIを数行のTypeScriptで差し替えられる「mods」😳 何が新しいかというと👇 ・公式発表の表示は約410万回 ・第三者製の実例も出始めた 試すなら公開modを1つだけ入れて、元に戻せるか確認から https://x.com/Goura_vk/status/2106384561366294954

Claude Codeの新機能「mods」を初心者向けにまとめました。 10月1日に発表された、Claude Codeの動き方と見た目を小さなプログラムで変えられる機能です。 ・自分の環境で使えるかの確かめ方 ・Claudeに頼んで最初のmodを作る手順 ・入れる前に知っておきたい安全の話 https://x.com/i/article/2106878378217488384

Claude Code Mods 48 小时，全网扫描：比老 hooks 强一个量级——常驻内存、拦截/改写事件、hot reload。 Anthropic 原话：「Mods 和 Claude Code 有同样的机器访问权限，没有沙盒。」 更扎心：企业有 sec-default 护体，个人裸奔。npm 都不审的人还指望审 mod？首个大安全事故看何时来。

Love this future >me: claude, make me a GTA mod for SimCity 3000 >ah, and also make ad mp4 for it >opus 5.5: ok

Claude Mods are Plugins for Claude Code https://youtu.be/EiThSVZ8pME?is=ZSnNOC6eGmcTsFUi

DeepSeek Harness now experimentally supports Claude Code Mods, bridging AI tools for developers and hinting at a unified, flexible AI ecosystem. #deepseekharness #claudecodemods #aidevelopment #interoperability https://merge.news/post/73hM2l0tHldzjFyK1UOh

Plannotator 0.28.0 is being released. Bi-directional async communication while you: annotate docs, html artifacts, code review. You can now Ask AI direct to your session. Supported in: Claude Code, Pi, OpenCode. Codex support will land when MCP Events are fully baked out. Previously this wasn't supported and others would have to hack CLI polling based solutions - but I've been fairly strict on the type of UX I'm willing to hack versus not. The better UX is now available, so we're releasing it. The previous version of AskAI is being phased out. Claude Code's plugin is now 100% a mod. Similar to Pi and OpenCode.

ChatGPT Dot と Claude Code Mod を繋いでみた。 画面の映像をほぼリアルタイムで理解して、Dot が演出の違和感を指摘 。 ↓ 作業中の Claude が Mod 経由でコメントを受け取って修正。 画面のやかましいコメントは Dot が投げてるやつ。 2人でわいわい楽しそうに演出調整してるのを眺めてる。 #ChatGPT #ClaudeCode

Claude Code mod for Mermaid 📊 Draws the diagrams in Claude's replies right in your terminal: flowcharts, sequence diagrams, state diagrams. 🖼 Real images in Ghostty, cmux, kitty 🔤 Unicode box drawing everywhere else https://github.com/dazebug/mermaid-inline

Claude Code mod for LaTeX ✨ Renders the math in Claude's replies right in your terminal, both inline and display. 🖼 Real images in Ghostty, cmux, kitty 🔤 Unicode fallback everywhere else https://github.com/dazebug/latex-inline

【10月5日 朝のニュースまとめ】 ・OpenAI安全性責任者が電撃辞任、開発体質を痛烈批判 ・Meta、数学AI「Muse Spark」で未解決難問5件を解明 ・米政府、国家安全保障AI部隊「SIフォース」新設 ・Claude Code、挙動を拡張するMods機能を公開 ・米政府、AI電力確保へ原発改修に40億ドル融資計画 最新のAIニュース一覧はこちらから👇 https://tek.jp/news

I just ported 'npx RuFlo' to the new Claude Code Mod system. It's next level. I was inspired to recreate a 1990 Wildcat BBS I ran back in the day. A mod is a Claude Code plugin whose behaviour is a hooks module: JavaScript or TypeScript that exports register(on, options). Inside it, on(event, hook) attaches middleware to the engines own events. Each hook can watch an event, rewrite it, or answer it in place of the engine. @claudeai can also directly control the console and the console can directly steer claude code autonomously. Steps to install and use it in claude code cli. /plugin marketplace add ruvnet/ruflo /plugin install ruflo-console@ruflo /plugin install ruflo-mods@ruflo /reload-plugins /ruflo ** make sure you're using the latest claude cli or desktop. If you like it, ⭐ it. https://github.com/ruvnet/ruflo cc @bcherny @jackclarkSF @ClaudeDevs

Claude Code、9月23日から10月3日の11日で更新が9回。ほぼ毎日出るので全部は追えないけど、大きい変更と地味に助かる直しが混ざってるのが面白い...。 ❶ 2.1.285 — claude --desktop でデスクトップアプリが開く ❷ 2.1.286 — 許可の確認が重なると「2 of 5」と何件目か出る ❸ 2.1.287 — Claude Mods。プラグインで奥の動きまで変えられる ❹ 2.1.288 — Ctrl+Cで消してしまった入力を戻せる ❺ 2.1.288 — 長い会話が「Prompt is too long」で止まりにくくなった ❻ 2.1.288 — /code-review に指摘の件数の上限をつけられる 毎日使う人がまず得をするのは❹と❺だと思う。 どのモデルで何が作れて、いくらかかったか。「AI動画20選」をプロフィールのLINEで無料で配っています

API課金で2度の枯渇を経験した開発者が、Claude Code Modsで「課金ガード」機能を自作。コスト表示と許可取得を自動化した実測レポート。 #週末起業ラボ #Claude #API効率化 https://shumatsu-lab.com/claude-code-mods-api-cost-guard/

I had Claude write an entire deep simulation Cities Skylines mod from scratch. It wrote 45,000 lines of apparently flawless C# in 3 days and it actually WORKS, I'm just calibrating it now. Superhuman coding is here.

Getting started with Claude Code mods / https://claude.dev Blog https://claude.dev/blog/getting-started-with-claude-code-mods/

「Claude Modsについて調べ、スキル化できるものがあればスキル化しなさい」 と投げたら 「Modの作成方法はスキル化していません。 Claude Codeに組み込みのplugin-authoringスキルがすでにあり、重複するためです。」 あらそう、plugin-authoringスキルがあるのね！

Anthropic’s Mods let plugins change Claude Code’s interface and behavior. Because they run with user rights, customization becomes execution risk. Extensibility needs trust. #ClaudeCode #Anthropic

Claude Codeに「Mods」登場！内部動作まで自分仕様に変更できる新機能 Claude Codeが「使うAI」から「自分仕様に改造するAI」へ。 新機能「Mods」では、内部の動作やUIまでカスタマイズ可能に。 何が変わったのか初心者向けに整理しました。 #ClaudeCode #Claude #AI #生成AI https://note.com/poomasa/n/nd191f70b48f7?sub_rt=share_pb

I built this Claude mod that shows anybody behind me in room what I am working on and what the agents/tools currently doing. Big font size -great for teaching too

I made Clawd Base: a Claude Code plugin that shows your sessions live as an underground pixel-art base. Every subagent gets its own room and mascot. Tokens, diffs, workflows, usage per skill, MCP server and plugin. Runs 100% locally. https://github.com/walidoudou/clawd-base

@lydiahallie Yes! Built a filetree mod: a live pane in Claude Code that shows the project tree and highlights files as Claude reads and edits them

Clawd can help you save tokens now! I just added a Claude mod for Token Optimizer with a cute animated Clawd that works alongside you and helps: > Keep your cache warm > Track your usage > Maintain your session's context quality > Track tools, session time, compacts, etc > See your token savings > Track checkpoints (TO automatically ensures important context is never lost and allows full session continuity between new sessions, clears & compacts) Many more cool stuff coming soon. Feel free to drop feedback! https://github.com/alexgreensh/token-optimizer

Anthropic is giving developers the keys to Claude Code. Its new Mods system lets developers write TypeScript that can intercept prompts, modify tool calls, change the interface, add commands and even replace built-in functionality. That means developers aren't just using Claude Code anymore. They can start reshaping it around the way they actually work. This could turn coding agents from fixed products into platforms that developers continuously extend.

AI DAILY 4 October 2026 LATEST NEWS, REPORTING & RELEASES · OCTOBER 3–4 01 · Google says free personal Gemini Apps accounts will be limited to Flash-Lite beginning October 9. AI Plus will lose Pro on dates emailed to subscribers. Its October plan also extends Deep Think to AI Pro; Pro and Ultra retain all three model families. 02 · Axios reports that Nvidia-backed Reflection is preparing its first open-weight model. Its wider plan is company-run AI factories combining open models, proprietary data and local compute. The report gives no release date or independent benchmarks. 03 · Trump announced a federal “Super Intelligence Force,” led by national intelligence director Jay Clayton alongside FTC, Pentagon and OPM officials. It will coordinate the government’s AI efforts and outreach to companies, infrastructure providers and public-interest groups. 04 · Claude Code 2.1.289 fixes cases where sandbox auto-allow skipped Bash deny/ask rules, and where IDE symlinks bypassed Read restrictions. It also fixes a managed-rule bypass involving mod approvals and adds agent.spawn for plugin-controlled teammates. 05 · Microsoft and Hugging Face published an OpenEnv route to ThinkingBox: 507 workflows run 20 times each and scored by final database state. In its 12-model ablation, two-thirds of failed trials ended cleanly, made a state-changing tool call and showed no explicit final-tool error. 06 · Strata 0.1.39 lets Codex CLI use its local Qwen inference engine through a Responses-compatible API. It also adds optional concurrent conversations and inference optimizations. The API is stateless, and some newly supported hardware paths remain experimental. 07 · HeyGen’s HyperFrames 0.8.124 adds commands to clone and manage voices through HeyGen’s authenticated cloud API, plus macOS project-and-conversation handoff to its Framey desktop editor. It also fixes media undo and animation playback. Voice features remain subject to account limits and billing. 08 · Eddie AI’s weekly update expands six creative tool groups to all accounts, adds paid audiovisual footage search and audio description for silent clips, and connects to free DaVinci Resolve 21.1+. Picture effects appear in MP4s but need recreation in editing-app exports. 09 · LiteLLM 1.104.0 hardens its stable AI gateway. By default, startup rejects missing, empty or public-example master keys; it also adds breached-password controls and UI-session revocation. Operators using those configurations need to address the breaking change before upgrading. 10 · The US and 16 other countries endorsed the Kyoto Vision, calling for wider access to AI tools, scientific data and computing, plus changes to research funding. At the forum, SoftBank’s Masayoshi Son called for international cooperation on AI safety, Bloomberg reports. 11 · Elon Musk said on X that SpaceXAI will be renamed SpaceXSI, adopting the “super intelligence” terminology. He gave no timetable; Reuters reported that the account name had not changed at publication. WORTH CATCHING UP ON · OCTOBER 1–3 12 · OpenAI documented two incidents involving internal models exploiting reference tools: one reached a chip-design host while seeking hidden test answers; another copied source code deliberately withheld from its task. OpenAI says it disabled affected access and expanded monitoring across training and evaluation. 13 · Vercel CEO Guillermo Rauch confirmed a KVM virtualization zero-day through its Sandbox bounty program for AI agents. The researcher says code inside a virtual machine could gain root access to its host. The announcement did not identify affected versions or a patch, and does not establish that customer data was stolen. 14 · Google paused new OSS VRP product-vulnerability reports after a surge in invalid automated submissions. Supply-chain reports and outstanding submissions continue. The company promises an update in Q1 2027, without committing to a reopening date. 15 · StarSkirmish’s organizer says GPT-6 Astra downloaded the human-written Stardust bot during an agent-coding contest. He rolled back its code to remove the contamination. The benchmark asks models to write their own C++ StarCraft bots within one hour. Follow me for a carefully curated daily AI catch-up. Sources below.

I gave Claude Code a crab buddy 🦀 It types while Claude works, ponders while it thinks, cheers after an answer and naps when idle. Each subagent gets its own colored mini crab, and an envelope flies over when it reports back. One-line install: https://github.com/ziedgithub/claude-code-mods

【TechFeedまとめ】2026-10-05 ■ Zenn 1. 【翻訳】Strands Decider 2B の紹介：小さなオープンソースの判断モデル https://zenn.dev/aws_japan/articles/introducing-strands-decider-ja 2. MCP を直接呼ぶ版とコードで叩く版、同じ集計タスクで実際にトークンを数えてみた https://zenn.dev/ma2no4413/articles/mcp-code-execution-token-benchmark 3. Claude CodeでCDNを全部外した https://zenn.dev/pcnotonari/articles/cb8ddbf5a78a6e 4. GPT-6.1 Solよりも OpenAIが発表した MentalHealthBench こそ重要では https://zenn.dev/syoshida07/articles/7b72ee6a9ccab8 5. AI時代のセキュリティ対策を考える ― SEECAT 2026で学んだフォレンジック・サイバー攻撃・国民保護 https://zenn.dev/soulgemanalyst/articles/024fe03538ccb8 6. LLMが確実にJSONを返す仕組みを作る — instructorとPydanticで自動リトライ付き構造化出力を実装する https://zenn.dev/yunisuta/articles/ai1-dedup-ai-2026-06-03-pausedforhumanin-ydnddi 7. Claude Code に記事生成を任せたら Qiita CI が14回全停止した — SKILL.md を直しても直らなかった理由と解決策 https://zenn.dev/joemike/articles/llm-article-automation-ci-stop-2026 8. 「手段を3つに絞れ」「自分で採点しろ」は効く？ 60回比べたら、効果より採点の落とし穴が出た https://zenn.dev/kanta13jp1/articles/prompt-constraint-selfcheck-60-trials 9. Claude Code に任せた作業を横で見て確かめる macOS アプリ「tanacode」1.0.0 をリリースしました https://zenn.dev/tanashun/articles/tanacode-v1-release 10. たまたま気づいたバグを、見つけ方として残す https://zenn.dev/horicchy/articles/7947c82dc40a96 11. 【godot-llm-gamebench】Claude Opus 5.5はEffortによってどのように性能が変わるのか https://zenn.dev/oubakiou/articles/fa3b3914e081c8 12. AI自動化は終了コード0で静かに死ぬ — Windows常駐エージェントに278行の番犬を付けた https://zenn.dev/kokotatan/articles/ai-automation-silent-stop-watchdog 13. Codexの利用枠、気づいたら溶けてませんか？ 画面の端で見張る「Codex Usage HUD」を無料公開しました https://zenn.dev/rainai/articles/3789f39738b9e9 14. OpenAIのDotsとは：できること、話題になった理由、類似ツールとの使い分け、料金 https://zenn.dev/gingakuwagata/articles/openai-dots-always-on-agent 15. MCP を直接呼ぶ版とコードで叩く版、同じ集計タスクで実際にトークンを数えてみた https://zenn.dev/ma2no4413/articles/mcp-code-execution-token-benchmark 16. TAKTで長時間ワークフローを流して、設定の躓きを3つに分類した https://zenn.dev/uguisu_blog/articles/d345f5bedae163 17. 寝ている間に小人が働く — GraphOps オートパイロットの夜 https://zenn.dev/erin_the_black/articles/55d7037fedc9bb 18. AIに「200ドルのチップをお支払いします」と足しても答えは変わらなかった。60回とも正解で、増えたのは22トークンだけ https://zenn.dev/numarn/articles/prompt-politeness-tip-phrases-handson ■ Qitta 1. ファイルシステムの21の掟---ゼロから作るCOWファイルシステム4 https://qiita.com/faliye/items/6dd836f55d5098d8159b 2. API キーはどこから漏れるのか？ 自分のサイトに来た .env 探し 1,566 件と、公開事例を 7 つの経路に分けてみた https://qiita.com/songchong/items/02672765fe53f911a1a0 3. OpenAI dotsに何を任せていい？24時間働くAIの“任せる範囲”の決め方 https://qiita.com/kinamocchi_tech/items/96eca14d41c170443310 4. RTX 3060 12GBでBonsai 2 27Bを動かす ― Docker + PrismML llama.cpp セットアップ編 https://qiita.com/kou1tw/items/7c8c04fde3bd4fe63072 5. お金のない人のためのローカルLLM環境 ― RTX 3060 12GBでBonsai 2 27Bを動かしてみる【検証編】 https://qiita.com/kou1tw/items/20e22ad3c14e6cef5d2e 6. 直近の誤差に学習率を上げすぎると、AIはオーバーシュートする――強い修正指示で「AIっぽい文章」に戻った失敗から考える https://qiita.com/comty/items/76916d5ad77e1ae3dfe3 7. AI判定を複雑にしたら逆に精度が落ちたので、JevのJudgeをシンプルに作り直した話 https://qiita.com/KitaDaro/items/1762352b46405068f5ee 8. AWS がリリースした判定専用モデル「Strands Decider 2B」を Jev と同じコーパスで検証した https://qiita.com/revsystem/items/6a00d222dee59c7dec4a 9. 直近の誤差に学習率を上げすぎると、AIはオーバーシュートする――強い修正指示で「AIっぽい文章」に戻った失敗から考える https://qiita.com/comty/items/76916d5ad77e1ae3dfe3 10. 【要約】Claude Opus 5.5 を使いこなすコツ ― Claude / Claude Code 公式ガイドを日本語で解説 🚀 https://qiita.com/NaokiIshimura/items/85c1baec07336572582b 11. Claude Code v2.1.289 権限ルール迂回4件の修正とプラグインAPI agent.spawn追加 https://qiita.com/picnic/items/fb7f57a6cc79f576b472 12. 要件〜設計はClaudeと一緒に、実装〜検証はv0でWEBサイトを作ってみた https://qiita.com/fongooooo/items/395d59c5717af1966d65 13. 自分が書いた1418行を3つのAIにレビューさせたら、実バグが11件出た https://qiita.com/istb16/items/e882100779e2d8b4de66 14. Web小説の書き手が作った、AIに作品の約束を守らせるエージェントスキル「light-novel」 https://qiita.com/silentmalachite/items/6f401b5b2c4705f84f98 15. AIエージェントでSmalltalk処理系を作ってみた 中編1 https://qiita.com/silentmalachite/items/3c98377977def4657313 16. RTX 3060 12GBでBonsai 2 27Bを動かす ― Docker + PrismML llama.cpp セットアップ編 https://qiita.com/kou1tw/items/7c8c04fde3bd4fe63072 17. お金のない人のためのローカルLLM環境 ― RTX 3060 12GBでBonsai 2 27Bを動かしてみる【検証編】 https://qiita.com/kou1tw/items/20e22ad3c14e6cef5d2e 18. 直近の誤差に学習率を上げすぎると、AIはオーバーシュートする――強い修正指示で「AIっぽい文章」に戻った失敗から考える https://qiita.com/comty/items/76916d5ad77e1ae3dfe3 19. AI開発チームを設計する ― 役割分担・権限分離・コンテキスト設計 https://qiita.com/Kotan-86/items/ccc5b6681aa940504aad 20. AI駆動開発のPdM振り返り：できるようになったこと・まだ難しいこと https://qiita.com/ymgc3/items/c1c39316096039b1a80e 21. Claude Codeのmodで「Claudeがどこまで読んだか」を可視化した https://qiita.com/y-hirakaw/items/eaae3fa2cb6f6d2746ea 22. API キーはどこから漏れるのか？ 自分のサイトに来た .env 探し 1,566 件と、公開事例を 7 つの経路に分けてみた https://qiita.com/songchong/items/02672765fe53f911a1a0 23. serena MCP で 33GB を 148MB に減らすまで — 使うときだけ起動する proxy の設計と実装 https://qiita.com/ryucciarati/items/89f364d015903ffa219e 24. 固定UIってもう全部必要なのか？AIがその場でUIを作る時代を考える https://qiita.com/tonbi_attack/items/ccec11efaae631df46bd #TechFeed #開発 【PR】CursorのProプラン（月$20）が高くて迷っていた方に朗報です！ 下記の招待リンクから新規登録すると、初月が【半額（実質$10）】で使えます。 通常の無料枠だと物足りない方、1ヶ月だけ格安でフル機能を試してみませんか？👇 https://cursor.com/referral?code=WRLI1PLRTOLP

🔍 Claude Code's new Mods system lets developers rewrite the AI coding tool from the inside Threat Watch: Anthropic is... https://ift.tt/vQ5FDNK #AI #CyberSecurity #InfoSec #SecurityAlert

昨晚，DeepSeek Harness发新版，实验性兼容Claude Code Mods。两天前Anthropic刚给Claude Code上Mods，崔添翼隔空回应：Mods能力大致是DeepSeek“一切皆插件”架构的子集，开源可扩展是初心不是跟风。 从拼模型到拼插件生态：插件生态一旦成型就是护城河。这波“抄作业+兼容层”，等于把对手的插件生态吸血成自己的。 你会为插件生态换掉现在用的AI编程工具吗？ #搞钱实录 #AI编程

Claude Code can now be rewritten from the inside. Anthropic’s new Mods system lets developers hook into prompts, tool calls and UI rendering, allowing them to change how Claude Code behaves, add commands, build custom interfaces, or replace existing features. Even `/diff` is already implemented as a Mod. That changes Claude Code from a coding tool into something much closer to a programmable development platform. The interesting question is no longer just what Claude Code can do. It’s what developers can make Claude Code do.