- AI Fire
- Posts
- 🔌 Google Backs Portable AI Agent Plugins
Your favorite AI agents may soon start sharing the same skills. Google, OpenAI, Microsoft, Amazon, Cursor, and Vercel are now backing a standard designed to make agent plugins portable across different platforms.
What's on FIRE 🔥
IN PARTNERSHIP WITH BELAY
When you love what you do, it can be easy to take on more — more tasks, more deadlines, more hours – but before you know it, you don’t have time to do what you loved in the beginning. Don’t just do more – do more of what you do best.
BELAY’s flexible staffing solutions leverage industry experience with AI systems to increase productivity without sacrificing quality. You can accomplish more and juggle less with our exceptional U.S.-based Virtual Assistants, Accounting Professionals, and Marketing Assistants. Learn how with our free ebook, Delegate to Elevate, and leave the more to BELAY.
AI INSIGHTS
Google just joined Agent Plugins as a Core Maintainer, backing a new open standard designed to make Agent Skills and MCP servers portable across different AI clients.
The standard is already supported by maintainers from Amazon, Cursor, Microsoft, OpenAI, Vercel, and Google.
The main problem is simple: developers can build a working Skill or MCP server, then still need to change folders, manifests, and configs for every new client.
Agent Plugins 1.0 fixes that with one shared package structure:
plugin.jsonstores the basic plugin information.skills/contains reusable Agent Skills.mcp.jsondefines connected MCP servers.Client-specific features can stay in their own extension folders.
That means a developer could package the same capability once and use it across tools like Gemini CLI, Claude Code, Cursor, and Antigravity without maintaining several separate versions.
Google is also keeping version 1 intentionally small. Agent Plugins currently doesn’t define installation, permissions, sandboxing, distribution, or security rules. Each client can handle those parts in its own way.
Google already supports the format in two products:
Agents CLI, which brings Google’s agent-building skills into multiple coding agents.
Data Agent Kit, which packages skills and MCP servers for services such as BigQuery, Spanner, and Cloud SQL.
This may sound like boring infrastructure, but it could solve a growing problem in the agent ecosystem. If more platforms adopt the same package format, developers can build useful agent extensions once instead of rebuilding the wrapper for every AI client.
PRESENTED BY MESH API
Know Exactly Who's Spending Your AI Budget.
Every AI request leaves a trail. Mesh gives engineering and finance complete visibility into who used which model, how many tokens were consumed and where your AI budget is going.
Stop guessing. Start governing AI spend.
Connect once, switch between GPT, Claude, Gemini and hundreds more whenever you want, while automatically routing requests for 40% lower costs and 99.99% AI response rate.
AI SOURCES FROM AI FIRE
1. 3 AI Video Generators That Are ACTUALLY FREE with Daily Credits (Almost Unlimited). Creating videos is NOT expensive at all with these 3 Free AI video tools. They still offer generous free daily credits. I tested how far you can actually go with their free plans.
2. This #1 Trending Claude Skill Gives Research a Serious Upgrade. Last 30 Days searches Reddit, X, YouTube, Hacker News, GitHub, and more to show what real users are saying, with 55K GitHub stars behind it.
3. Opus 5’s New Prompting Guide: 5 Old Claude Rules You Should Change. Learn how Anthropic recommends prompting Opus 5, from choosing the right effort level to setting clear limits so Claude gives better results without wasting tokens.
FIRE RECAP: BIGGEST AI NEWS THIS WEEK
🆓 OpenAI made GPT-5.6 Luna the default for Free & Go users, with unlimited text chats and a new Think button. Plus and Pro users also got an updated GPT-5.6 Sol.
🧠 Google just reshuffled its AI leadership. Demis Hassabis is now Chair of Google DeepMind & Alphabet’s Chief Scientist, while Koray Kavukcuoglu takes over Gemini development and frontier AI research.
🤖 Claude Code will make Auto Mode the default for new Pro, Max & Team sessions starting August 14. Anthropic says it catches 89% of dangerous commands, compared with just 13.6% for humans.
📊 OpenAI released its first country-by-country look at how people use ChatGPT. Work users are over 2× more likely to use it to create or finish something, while multimedia is now its fastest-growing use case.
🏢 Anthropic now lets Team & Enterprise companies run Claude Code on their own infrastructure. Code, secrets & files stay inside company-controlled systems while Claude handles model inference.
TODAY IN AI
AI HIGHLIGHTS
🛡️ OpenAI is slowing work on its upcoming Astra model after tests showed it may have “critical” cyber capabilities. OpenAI says it needs stronger safeguards before any release.
🌐 Cloudflare just launched Kitesurf, a browser built specifically for AI agents. It uses up to 7× less memory than Chromium for some tasks and is free to try in beta through Browser Run.
💬 Claude Code sessions can now talk to each other on macOS & Linux. One session can share findings, warn another about broken changes, or send answers that help unblock work on the same project.
🎵 Suno plans to add invisible watermarks to AI-generated music and limit some downloads. The goal is to help platforms detect AI tracks and reduce large-scale abuse as copyright pressure keeps growing.
😳 Google’s AI somehow revealed an unreleased game character name that existed only in a private Google Doc. Google says private Workspace files aren’t used to train Gemini.
💰 AI M&A: OpenAI acquired NextSlide, an AI startup that turns prompts, notes, documents, and research into polished, editable presentations. The NextSlide team has now joined OpenAI to help build ChatGPT, bringing its AI presentation tech into a much larger product ecosystem.
NEW EMPOWERED AI TOOLS
📊 Prefactor scores AI agent runs in real time, helping teams catch quality regressions, drift, and production issues before they scale.
🎙️ SKI brings free voice coding to Claude Code, Codex, and more, letting you talk with your coding agent like a teammate on Mac or Windows.
📣 Adomate turns data from Meta, Trustpilot, Amazon, and ad libraries into traceable, branded ad concepts at scale.
🧠 Memmy Agent gives Claude Code, Codex, OpenClaw, and Hermes one shared local memory for chats, preferences, decisions, and ongoing work.
AI BREAKTHROUGH
Google DeepMind’s WeatherNext Cyclones can predict a storm’s track, intensity, and wind structure with more than 24 hours of extra lead time.
The key results:
Its 3-day forecast matches the accuracy older models reached at 2 days, roughly equal to 10 years of forecasting progress.
Google trained WeatherNext on nearly 20TB of atmospheric data plus almost 5,000 historical storms.
Unlike traditional systems that often separate track and intensity forecasting, one AI model handles both.
Its Functional Generative Networks (FGNs) can generate 1,000 possible storm scenarios, including rare rapid-intensification events.
WeatherNext can produce a 15-day forecast in under one minute on a TPU, using data around 100x coarser than traditional high-resolution models.
The model already helped the National Hurricane Center forecast Hurricane Melissa’s rapid intensification and Jamaica landfall in 2025.
Google is now open sourcing WeatherNext Cyclones, WeatherNext 2, and WeatherNext 2-mini, giving researchers and weather agencies access to the models behind the breakthrough.
We read your emails, comments, and poll replies daily
How would you rate today’s newsletter?Your feedback helps us create the best newsletter possible |
Hit reply and say Hello – we'd love to hear from you!
Like what you're reading? Forward it to friends, and they can sign up here.
Cheers,
The AI Fire Team
Reply