I'm proud to be a 2026 AAIF Community Organizer for Luxembourg. Building vendor-neutral, builder-first nights for agentic AI — deep dives, live demos, and shipping agents. Let's connect: https://t.co/wYCYacwN5V
https://t.co/tzPbLXLnZW · @AgenticAIFdn#AAIFCommunity
@AgenticAIFdn Treat AGENTS.md like page one of your cookbook and skills like recipes opened on demand.
In fmind/dot (01f21ef), 77 lines (2,342 tokens) + 71 skill index entries (2,547) fit a 5k-token cap vs 66,523 on-demand tokens. #AIAgents
Your coding agent guesses your test runner and repo rules wrong? Customize your agents with a lean AGENTS.md, written like onboarding a teammate on day one.
AI Agents in a Nut$SHELL, episode 6: why auto-generating this @AgenticAIFdn standard doesn't pay off.
@AgenticAIFdn Split two complementary files:
- Personal ~/.agents/AGENTS.md for collaboration rules, symlinked to each CLI
- Repo AGENTS.md for test commands and traps (import in CLAUDE.md if present)
Write from your head, then ask the agent to compress.
I did say we'd be releasing our models periodically...introducing Gemini 4 Carbon, our next frontier model.
Following Argon's release just two weeks ago, Carbon shows the pace at which we’re pushing the frontier, it ranks #1 on the recently updated v5.0 of @ArtificialAnlys Intelligence Index at 59. It also leads across key cybersecurity evaluations scoring 88.2% on SEC-bench Pro for finding vulnerabilities, 54.8% on ExploitGym for exploitation, and 71.0% on CWE-bench for patching.
We’re seeing significant improvements in long-horizon tasks even compared to Argon, and we’ve strengthened frontier safeguards.
As we did with Argon, we’re going to roll it out gradually + responsibly. Some trusted cyber defenders will get access starting this week. We'll be expanding it to more partners through our Fairwind Program, ahead of wider availability soon, so hang tight!
Huge thanks to our host @awscloud, our local chapter supporters @PwC_Luxembourg, @ArcelorMittal_L, @SnT_lu, @Microsoft & @Google, my fellow co-organizers (Mustafa Aldemir, Hazal Kantarci, Hajar Khizou, Maxime Cordy & Pablo Fernández de Diego), and everyone who packed the room!
Rocking event at the @awscloud Lab for our 2nd @AgenticAIFdn (AAIF) #Luxembourg night!
1/ Live-coded an agent in 7 steps from a local model on a Mac to AWS Bedrock, model routing, AGENTS.md & MCP
2/ Live voice agent+ agentic robotic arm on stage
Chapter: https://t.co/wYCYacwN5V
Huge thanks to our host @awscloud, our local chapter supporters @PwC_Luxembourg, @ArcelorMittal_L, @SnT_lu, @Microsoft & @Google, my fellow co-organizers (Mustafa Aldemir, Hazal Kantarci, Hajar Khizou & Pablo Fernández de Diego), and everyone who packed the room!
Two routes: a documented issue via the project's contribution instructions, or a public working group for cross-project problems.
You may not get the change you wanted. What you should get is a decision you can inspect.
Two companies push conflicting changes to the same open-source agent component. Your production system depends on it. What do you do?
AI Agents in a Nut$SHELL: the Agentic AI Foundation (AAIF) in under 4 minutes. https://t.co/A8JI7Gz4wW
I'm also considering frontier open-weight models on OpenRouter. For now, I prefer the proprietary agent tools: I find them more capable out of the box.
What are you running? One subscription, several, or mostly APIs?
Claude Opus 5.5 and Sonnet 5.5 are now in @antigravity !
Meanwhile, I'm trimming my AI subscriptions: cancelling my $200/month ChatGPT plan and keeping Claude + Google AI Ultra. Here's why.
Google AI Ultra also lets me share eligible AI and storage benefits with my wife, with 30 TB across Drive, Gmail and Photos, plus $100/month in GCP credits for me.
The value goes well beyond the coding tools.
My dotfiles 2.0 now have two users: me and my coding agents.
Unix gives us files, commands and pipes. AGENTS.md explains how to leverage the system.
Better dotfiles help both of us. Agents can use the same toolkit as humans.
Find more in my new article:
https://t.co/BwGH3crcKi
Packed room last night for the very first @AgenticAI_org #Luxembourg gathering at @SnT_lu!
Zero vendor hype, just builders, researchers, and enterprise leads sharing what it actually takes to make #AI#agents reliable in production.
A few moments from the evening below 👇
Nothing quite like watching your loop iterate on multi-step tool attacks and climb the leaderboard while you sleep or focus on daily work.
Autonomous loops deliver real results when you steer them right. Next goal: tuning the system for gold.
Took 69th place out of 4,251 teams (Silver Medal) in #Kaggle's AI Agent Security competition—powered entirely by an #AI#agent#loop.
I wanted to enter Kaggle but lacked time. So I built an agent loop to compete for me while I steered.
https://t.co/UjVzQZnUbP