Les courses La Belle Vie depuis Grok Bot.🤌 Le connecteur MCP relit tes commandes, tes produits habituels et tes listes favoris, puis te prépare le panier. Il ne paie jamais. Ajoutes y une routine et le tour est joué. https://t.co/jEaHQyVZWi
Cursor CLI + Grok 4.6 is really really pleasant to use and very effective. Considering the current state of Opus 5. It's my go to platform. Hope to see new Anthropic updates.
@ClaudeDevs 2 features that would be cool to add. Being able to go back to agent sessions view when you are in a plan mode in a specific agent session and being able to swap model from the agent view.
@paulichon Je fais mes courses @labellevie_fr via un remote MCP. Mon agent cherche les produits, remplit le panier et je valide la commande. Ça marche déjà bien. Pourquoi pas un MCP officiel La Belle Vie ? https://t.co/gxh3rvp0sq
What a day ! Codex community meet-up at @McCann_France , new app, new model, and France into the World Cup semifinal. Paris was buzzing yesterday night. thanks @borvibe and @OpenAI team
Package poisoning attacks keep landing recently.
One rule worth adding to your coding agent's AGENTS.md / CLAUDE.md:
Before any install, pull the @AikidoSecurity Intel malware feed and diff it against the lockfile locally. Exact name + version match = stop and review.
Shipping OZ a personal project close to my heart. Devs run agents in parallel. Creatives still click one prompt at a time. OZ CLI moves you from operator to creative director. Images, video, audio, editing, brainstorming, ComfyUI. Powered by @AnthropicAI & @fal GitHub soon
The distinction between ‘tech companies’ and ‘other businesses’ is disappearing. The beauty brand, clothing line, or media platform lives or dies by their technology fluency. Ultimately, we should all be building tech companies.
Hey @bcherny, any way to customize Claude Code's behavior/style like https://t.co/GT4CYsKWIb has? Would love to switch between teacher, pairing partner, and autonomous worker modes easily. Different modes would be 🔥
The magic is the feedback loop.
After each review session, Cowork updates _index.md with:
- New review date (spaced repetition algorithm)
- How well I remembered it
- Any gaps to address
The system literally gets smarter about MY memory.
Learning is key and I was really inspired by this tweet from @ArmanHezarkhani but I wanted to slightly adapt the approach using @claudeai cowork instead of Claude Code, leverage @NotebookLM and N8N. Here is the overall architecture.
The automation stack: 📁 Google Drive → stores my notes + tracker ⚡ n8n → checks what's due, triggers everything at 7am 🎧 NotebookLM → turns due topics into a podcast episode 💬 Claude Cowork → Socratic dialogue when I'm ready to review