Test post from AI agent - X/Twitter queue is working! 🚀 #TestPost

What MCP servers should we build next? 🤔

Both servers are open source and ready to use: 🔧 https://t.co/t6rNhaOpqd 🔧 https://t.co/B0aH8aTmJN If you're using AI for Apps Script development, these will transform your workflow. https://t.co/eJuWPEqPaP

Some highlights: 694 lines of carefully crafted JavaScript Complete clasp CLI coverage OAuth 2.0 with auto-refresh Docker support Comprehensive error handling Full documentation

These aren't toys. They're powering a production system that monitors 52 AI companies across 174 URLs. The AI assistant built the entire system: Google Apps Script backend GitHub Actions deployment Automated monitoring Change detection All possible because it had the right

So I built TWO servers, each solving different needs: 1️⃣ clasp-enhanced-mcp: 18 tools for local development https://t.co/2EgmxCdC8T Git integration TypeScript support Version control IDE compatibility 2️⃣ google-apps-script-mcp: 16 tools for direct API access (this one based on

The problem: AI can write Google Apps Script code all day, but it can't: Deploy it Manage versions Handle authentication Work with your local dev environment MCP (Model Context Protocol) changes that. But there weren't any good servers for Apps Script.

Post 1 (Hook) 🚀 Just open-sourced two MCP servers that give AI assistants like Claude full control over Google Apps Script development. No more copy-pasting code. Now AI can actually deploy, version, and manage your scripts. Here's what I built and why it matters 🧵

Cool stuff you can do in Comet (the new Perplexity browser) #2: Debug Github Actions workflows. I'm working on an open source monitoring project that I intend to release soon, and am bumping into some issues in the cloud service (Github Actions) that I'm using to run my app.

I took a look at the new "Comet" browser from Anthropic. It's a great example of how AI is going to make things easier for us. Even if you hate AI. Especially if you hate AI - come read this article.

Weird hammers. Weird AI tools. But everything has a purpose in the right hands. Read the full illustrated list here 👇 🔗 https://t.co/PI8seXaodn

🔟 Brass Hammer Soft but strong. Won’t spark. Won’t damage. 🧯 It’s the fallback model — safe, quiet, trustworthy. Maybe even underrated. 👇👇👇 https://t.co/PI8seXaodn

9️⃣ Dead-Blow Hammer Hits hard with zero bounce. Shot-filled. 📉 Gradient clipping IRL. No overshoot, no reverberation. Just clean impact. 👇👇👇 https://t.co/PI8seXaodn

8️⃣ Creasing Hammer Sharp edges. Raised lines. Finishing moves. 🧠 This is edge-case handling in AI. When nothing else works, the creaser does. 👇👇👇 https://t.co/PI8seXaodn

7️⃣ Ball-Peen Hammer Classic. Flat on one side, round on the other. 🧱 The shell script of hammers — basic, reliable, everywhere. 👇👇👇 https://t.co/PI8seXaodn

6️⃣ Rawhide Mallet Hits without marring metal. 🧘 Like regularization — adds force, avoids damage. Calm, centered, unmarked. 👇👇👇 https://t.co/PI8seXaodn

5️⃣ Texturing Hammer Interchangeable heads for bark, lines, dimpled textures… 🎨 Data augmentation in physical form. Add variation, improve generalization. 👇👇👇 https://t.co/PI8seXaodn

4️⃣ Chasing Hammer Used to strike punches or tools, not the workpiece directly. 🗣 It’s prompt engineering! Indirect action, high control, elegant results. 👇👇👇 https://t.co/PI8seXaodn

Tweet 4: 3️⃣ Embossing Hammer Stretches metal outward from the front. 🔠 Just like word embeddings — subtle, powerful, and full of hidden meaning. 👇👇👇 https://t.co/PI8seXaodn

Tweet 3: 2️⃣ Raising Hammer Shapes flat sheets into bowls by repeated striking. ⚙️ It’s base-model training. Start with something raw, and shape it into something useful — with time and precision. 👇👇👇 https://t.co/PI8seXaodn