Community posts

Claude Code mods let you hook into Claude's events and draw your own panels in the terminal. Mine watches subagents and charts them live: tools, timing, and the spawn tree. #ClaudeCode #ClaudeMods @claudeai

While everyone was exploring Claude Mods inside Claude, I started figuring out how to integrate them into CrewTower. It’s already working. CrewTower now gets clearer agent data and offers more actions without opening the terminal or another app. Mods might be exactly what I was waiting for. CrewTower lets you monitor and control AI coding agents from your MacBook notch.

@jarrodwatts Peak agent economy: humans fragging each other while the actual worker thinks. Next mod idea, a market on whether Claude finishes before you respawn. Our 5,580 test agents would quote it at 85% confidence and be right about half the time, which feels on-brand.

Anthropic shipped Claude Code mods: small TypeScript functions that can rewrite prompts, tool calls, or the UI, and they ship inside plugins. Some built-ins like /diff are mods now. Feels like Claude Code just grew a plugin API for its own brain

Speak That is now a universal plugin & skill that works across: • Google Antigravity (agy): agy plugin install ./plugins/speak-that • Claude Code: npx skills add Curtismith187/grok-build-speak-that --skill speak-that -g -y -a claude-code • Gemini CLI: npx skills add Curtismith187/grok-build-speak-that --skill speak-that -g -y -a gemini-cli • Codex: npx skills add Curtismith187/grok-build-speak-that --skill speak-that -g -y -a codex • Grok Build (where it all started!): Copy to ~/.grok/skills/ or run the 1-line installer • Plain PowerShell & Bash (direct pipe): Get-Content reply.md | speak-that cat reply.md | speak-that Originally built this specifically for Grok Build, but adapted & expanded it so everyone across all major AI coding agents can use it. Just type or dictate: 👉 "speak that" (or /speak) It streams the assistant’s last reply aloud verbatim via @xAI TTS directly to your headset while you stay in the flow. ⚠️ Bring your own xAI API key: Practice basic dev hygiene — huwag ilagay sa chat at lalong huwag i-post online! Set it via environment variable (XAI_API_KEY) or in your local .env / ~/.speak-that/config.json. Keep your keys secure. 📦 1-line full installer: • Windows: irm https://raw.githubusercontent.com/Curtismith187/grok-build-speak-that/main/install.ps1 | iex • Mac/Linux: curl -fsSL https://raw.githubusercontent.com/Curtismith187/grok-build-speak-that/main/install.sh | bash GitHub (MIT): https://github.com/Curtismith187/grok-build-speak-that @elonmusk @build @bot @SpaceXAI @cursor_ai

Matt Wolfe's weekly AI news roundup just dropped — and this week is one of the busiest in months. OpenAI's DevDay 2026: Dots, GPT-6.1 Sol, a Decisions API, ChatGPT Space and Pages, a meeting plugin, and Sign in with ChatGPT. Anthropic fired back with Claude Sonnet 5.5 and Claude Code Mods. Google shipped Gemini 4 Argon. xAI launched Team Bots — and used the moment to troll OpenAI's Dots launch. Also covered: Ideogram 4.5, Strands Decider 2B, ElevenLabs Eleven v4, Microsoft's MAI-Voice-2.1 and a streaming transcription model, plus a Trump superintelligence executive order. 27 minutes, every release timestamped so you can skip to what matters to you. If you build with AI or just want to stay current, this is the fastest way to catch up on the whole week. Follow @mreflow for the weekly drops. #AINews https://youtu.be/dDgncbBAA0c

Claude Code Mods push plugins deeper into the agent lifecycle: they can change behavior, intercept tool calls and customize the UI. Powerful, but it raises the trust bar. Once extensions can reshape permissions and execution, provenance matters as much as capability.

Claude Code mods in 4 steps (TypeScript or ask Claude to write one): - A mod is a small TypeScript function that rewrites prompts, adds UI, or replaces a built-in - Ship it inside a plugin (same package as skills/hooks/MCP) - Install with /plugin in the CLI or desktop app (or: claude plugin install name@marketplace) - Not sandboxed: only install from sources you trust Built-in /diff is already a mod you can turn off or replace. Which would you ship first: a usage cut-off, remote control, or a custom UI pane?

https://github.com/davidurco/cc-tamagotchi Tamagothchi as Claude Mod!!!!

Claude Code now has mods. Track context usage. Review commands before they run. Step through file edits. We explored three official Anthropic demos and explained where plugins and MCP fit. Watch the demos ↓ https://heymark.ai/en/guides/claude-code-mods-interface-plugins/

Custom cards are now in Claude Code with the Ottopus plugin powered by mods: watch the status of every plan, live, without asking the agent. This is pretty sick tbh.

readless is free and open source. Claude Code plugin that makes your agent speak when it’s stuck or done — so you can walk away without missing the moment. Install: /plugin marketplace add hwjustin/readless /plugin install readless /readless:setup GitHub: https://github.com/hwjustin/readless PyPI: https://pypi.org/project/readless-mcp/

Claude Code Mods are a pretty sick way to enhance the experience, especially in the CLI. I added an @ottopusxyz card that shows the live status of every plan, from the moment it's created to the onchain trade or transaction confirming. Hooks are an amazing way to feed that back into the UI.

@lydiahallie haven't yet, but 'middleware for claude code' is the framing that finally made the concept click for me. first one i'd build is a deploy-checklist mod — lint, changelog, zip, done. basically the boring release steps i always skip at 2am.

@jarrodwatts Yeah sorry mate should’ve linked your original sooner, my bad! Did it for a laugh don’t usually get much steam on my posts. https://x.com/alexchristou_/status/2106097986833174591?s=20

Og cred to this guy he actually built it into Claude code. Give it a try 👇 https://x.com/jarrodwatts/status/2105858869482471602?s=20

The special thing about this mod is that Claude can play the game for you. It captures frames from the screen and takes actions. Would be interesting to see what kind of games you can build with these mods that Claude can play against other AI agents.

@Nixtrodamis you can use mods in the cli as well! it's both desktop and cli

@bcherny Can a mod show a live diagram of the code Claude is changing? I keep asking it for artifacts to understand stuff. Would love that in the coding UI

Mods are out and you still do not want to hand-write TypeScript for every behavior change. The platform is there. The missing piece is a skill that helps Claude draft the Mod for you. claude-mod-builder is a free skill that helps Claude write Claude Mods. Install it, describe the behavior you want, and let the agent draft the Mod instead of starting from a blank file. When Mods are available, start with the skill that writes them with you.

@danieldeusing capi-cc-mod (MIT Claude Code mod that quizzes you on a language above the prompt while Claude thinks) fits awesome-ai-plugins Development & Workflow. Open the PR yourself: https://github.com/hashgraph-online/awesome-ai-plugins/blob/main/CONTRIBUTING.md

anthropic on claude code mods. truely love ya, but also come on...

Doom runs on anything. Now it runs inside Claude Code. I built a mod so you can now play Doom inside your Claude Code terminal. When you are busy Claude plays for you now.

My reel editor is now an open-source Claude Code plugin marketplace. How it works: You give it any video where you're talking. It hands back a finished reel with word-timed captions, animated scenes that land exactly when you say each point, music that ducks under your voice, sound effects, and an Instagram-ready export under 30 MB. Under the hood: → whisper.cpp transcribes your audio → Your exact transcript is matched to the audio word by word, even for Hinglish → Claude plans the scenes on a timeline and writes the edit as a React (Remotion) composition → It shows you preview frames before the full render → FFmpeg mixes the music and sound effects and does the final export → It checks frames from the final file before handing it over Plugins: The engine is one plugin, and every look is a separate theme plugin. Three themes ship today, one from each of my recent reels: → yellow-ticker: yellow card, swipe-in scenes, ticker captions → lime-bento: dark grid, bento layout, speaker splits into shards → cream-passport: paper documents, rubber stamps, ID cards Want your own style? A theme is just a kit.tsx (layout, captions, components), a http://sound.py (music and sound effects in pure numpy) and one example reel. There's a make-theme skill that walks you through building one. Install: Open your terminal, type claude to start Claude Code, then run: /plugin marketplace add srvjha/reel-animated-explainer /plugin install reel-studio@reel-animated-explainer /plugin install theme-yellow-ticker@reel-animated-explainer Then: /reel-studio:edit-reel my-video.mp4 It runs on your own Claude Code login, so there's nothing extra to connect. Repo: http://github.com/srvjha/reel-animated-explainer If anyone wants to contribute by adding a theme, PRs are welcome. Build one, and I’ll add it to the marketplace 🙂 Thanks to @pantharshit007 for this idea 😊

🔴 Right now on stream: Claude Code on Opus builds a mod for Skyrim × Fortnite. Build 11: "Loot, Build and the Gold Scar". Next mix vote is open: burn $SIGF to pick the two games. It closes when the AI finishes this build.

@bcherny Mods are amazing, Since this was in alpha a month back, I created mod which will mask secrets with similar strings locally, so none of the credentials will make it to the API servers of Anthropic or any other https://github.com/pratikbin/opensecretmask

4 anos no Brasil, portuguese still meh. so I built a Claude Code mod: a capybara quizzes me while Claude thinks 🦫 https://danieldeusing.de/articles/capi-claude-code-mod here since the early beta, gave feedback along the way, love where it's headed. keep it up @trq212 @bcherny and the whole @ClaudeDevs team

Codebase Mapper is the first plugin I install in every Claude Code project. A compact repo index at session start. Focused docs when Claude needs detail. The diagram shows a real command-to-code route in haiku-jar. Free + MIT: https://github.com/Eigenwise/eigenwise-toolshed/tree/main/plugins/codebase-mapper

@facilabo token-weather-usage (Apache-2.0 Claude Code mod: a line above the prompt with 5h/7d limit gauges, context weather, per-prompt token bar) fits awesome-ai-plugins Development & Workflow. Open the PR yourself: https://github.com/hashgraph-online/awesome-ai-plugins/blob/main/CONTRIBUTING.md

how to use claude code mods like a top 1% user, step by step: give this to your agent before everyone catches on👇 1. set up Jev connect Jev to the model registry you want to use. check that the connection works and the listed models are available. 2. build your claude code mod open claude code 2.1.287 or later and paste this prompt: “build a mod called run-ledger. load plugin-authoring and use the API types for my installed version. create a dashboard that shows: - estimated cost per run, model, and source plugin where known - which model handles each task - each subagent’s status, latest action, and elapsed time include token counts, cache usage, and reported retries. count each request once. keep background tasks linked to their original run. use dated prices. label costs as API estimates, not subscription charges. show unknown when data is missing. connect the mod to my existing Jev setup. give Jev the task, available models, prices, budget, and relevant past results. ask it to recommend a model and explain why. start with recommendations. make automatic routing optional for eligible subagents. show the recommended model, actual model, result, and cost. include Jev’s own cost. keep state across hot reloads. add details and export. the dashboard itself must make no model calls. validate the plugin. test rendering, costs, attribution, and duplicate counting. give me steps for a live test.” 3. test the setup allow hot reload when prompted. run a simple task, a subagent task, and a mod-triggered model call. check that the dashboard updates, each request counts once, and Jev’s recommended model can actually run the task. 4. install the working mod ask claude to copy it out of the temporary folder and install it as a persistent plugin. see the cost. choose the model. check the result.