Médéric Hurier (Fmind)

AI Architect (PhD) | AAIF Ambassador (LU) | Professional Cloud Architect (GCP) | Expertise: AI Agents, MLOps, Security

[Médéric Hurier (Fmind) @fmind_dev]

After countless nights of coding, I'm super excited to share my latest project: the MLOps Python Package v1.0.0 is now live on GitHub! 🚀

https://t.co/xsvLeFTYwv

#MLOps #Python #AI #ML #opensource

[Médéric Hurier (Fmind) @fmind_dev]

Short on time? Agent Payments Protocol (AP2) in a nutshell.

This video explains how Checkout and Payment Mandates create a verifiable authorization layer for autonomous agents without modifying underlying settlement rails.

#AI #Agents #AP2 #QuickLearn #BleedingAgent

[Médéric Hurier (Fmind) @fmind_dev]

Agent Payments Protocol (#AP2) on one page. 📌

Save it for your next #Agentic #Commerce design.

[Médéric Hurier (Fmind) @fmind_dev]

To ensure version stability, features now transit through a 12-month deprecation lifecycle. SDK betas for Python, TypeScript, Go, and C# are available ahead of the July 28 finalization.

Read the full analysis: https://t.co/Vuoi0bYSJN

[Médéric Hurier (Fmind) @fmind_dev]

The July 28, 2026 Model Context Protocol (MCP) spec is finalizing this month, bringing stateless scaling, enterprise authorization, and version stability to production deployments. Here's what is changing 🧵 #MCP #AIAgents #OpenStandards

[Médéric Hurier (Fmind) @fmind_dev]

Enterprise-Managed Authorization went stable in June. Uses token exchange (RFC 8693) to swap SSO identity for a per-server access token, enabling centralized policy and audit logs. Opt-in only. #CloudArchitecture

[Médéric Hurier (Fmind) @fmind_dev]

I was accepted as an official @AgenticAIFdn Ambassador! I'm super excited to help the Linux Foundation standardize #Agentic #AI (MCP, AGENTS.md, agentgateway, ...).

Program: https://t.co/lYt8PPczE6
Badge: https://t.co/D0ImDKKQGw

[Médéric Hurier (Fmind) @fmind_dev]

AI usage-based billing can be a budget killer. A client's bill spiked 5x when we transitioned to the new GitHub Copilot credit-based system.

Here is a FinOps playbook to compare LLM API vs. Self-Hosting (K8S) vs. Local Hosting (Mac OS): https://t.co/EDeeO55bcp #coding #agents

[Médéric Hurier (Fmind) @fmind_dev]

@rodydavis @antigravity A bundle between Google Workspace, Gemini Enterprise and Antigravity, so every teammate could benefit from the solution securely and according to their skill level.

Having to switch between subscriptions and preventing data collection is a pain point in entreprise settings.

[Médéric Hurier (Fmind) @fmind_dev]

Stop guessing software designs. I built agent-evolutions: genetic search for #coding #agents. It seeds, mutates, and scores variants on disk to find the best design without breaking a sweat (for you at least).

https://t.co/V16amSxunL

[Médéric Hurier (Fmind) @fmind_dev]

Too much time is spent babysitting #AI #Coding #Agents. We must minimize human attention while maximizing agent autonomy.

I built agent-levers: a set of agent skills running a Plan-Do-Check-Act loop for Claude, Antigravity, and Copilot.

https://t.co/HjAsoEGVA4

[Médéric Hurier (Fmind) @fmind_dev]

Coding agents retrieve wrong docs as often as they hallucinate.

Here come my new projects: #agent-#docs: a local reference for #Claude, #Gemini, #Copilot.

https://t.co/b03va5aOdY

[Médéric Hurier (Fmind) @fmind_dev]

Need a quick catch-up on Agent Skills? My new Agent in a Nut$SHELL channel keeps it fast!

Visualize the Agent Skills runtime in minutes.

https://t.co/FwjMOLnO7H

#AI #Agents #AgentSkills #NutShell

[Médéric Hurier (Fmind) @fmind_dev]

🚀 Launching Ep 3 of The Bleeding Agent Podcast

No hype, just deep technical dives into Agentic AI.

Episode 3: Agent Skills

Synthesizing key tech concepts with AI to help you build.

https://t.co/EJUVsu5atd

#AI #Agents #AgentSkills #DayOne

[Médéric Hurier (Fmind) @fmind_dev]

Tired of maintaining the same #subagent persona across Claude Code, Gemini CLI, Copilot, and Cursor?

I built #supagents: one source, compiled to every coding agent.

https://t.co/e5uGGSXuuK

[Médéric Hurier (Fmind) @fmind_dev]

We are rediscovering the Linux Philosophy for AI: handing an agent a raw CLI is faster, lighter, and more composable.

Read why: https://t.co/NhGRvALevS

#AI #GenAI #Linux #CLI

[Médéric Hurier (Fmind) @fmind_dev]

The hardest part of #AIcoding isn't the code, it's the context. Stop "#vibecoding" and steer your AI agents with the 5xP Framework: #Product, #Platform, #Process, #Profile, #Principle.

How to fix your AI context: https://t.co/6srQN064Ux

[Médéric Hurier (Fmind) @fmind_dev]

Need a quick catch-up on MCP? My new Agent in a Nut$SHELL channel keeps it fast!

Visualize the #Model #Context #Protocol (MCP) in minutes.

👇 https://t.co/0zoQyIMHwK

[Médéric Hurier (Fmind) @fmind_dev]

A visual guide to the Model Context Protocol (MCP).

If you've already caught our recent podcast deep dive, this infographic is your perfect cheat sheet. It maps out the architecture for standardizing AI context securely.

#AI #Agents #MCP #Infographic #Architecture