Welcome to Tech Updates: #9.
This Newsletter is a deep dive into the tools and AI workflows shaping the digital frontier. I’m sharing my process of building and exploring in the open so we can navigate the future of tech together.You can read the updates below, or watch the video version on YouTube. Embedded here for your enjoyment!
Research and Learning:
Learning Minimax
Reading about Google Chrome browser mcp.
Chrome for Developers@ChromiumDevWebMCP is available for early preview → goo.gle/4rML2O9 WebMCP aims to provide a standard way for exposing structured tools, ensuring AI agents can perform actions on your side with increased speed, reliability, and precision.
5:31 PM · Feb 13, 2026 · 1.91M Views182 Replies · 865 Reposts · 5.76K LikesReading about database MCP toolbox.
Akshay 🚀@akshay_pachaarGoogle open-sourced MCP Toolbox for Databases. I gave it access to everything else. For context, Google's MCP Toolbox for Databases is an open-source server that lets AI agents securely query structured databases like PostgreSQL and MySQL through the MCP protocol However, most12:30 PM · Feb 15, 2026 · 24K Views18 Replies · 38 Reposts · 275 LikesLearning picoclaw
Future Research to do:
migrating websites to Hugo + CDN.
Get IronClaw set up in a vm.
Security scores of AI models.
AISecHub@AISecHubAnthropic reveals its cybersecurity domination strategy. Claude and Gemini scored equally on cybersecurity tasks, but Claude Code scaffold showed its superiority. wiz.io/cyber-model-ar… Wiz just launched the AI Cyber Model Arena. 257 real-world challenges across five offensive
6:05 PM · Feb 14, 2026 · 6.39K Views4 Replies · 28 Reposts · 116 Likes
commands to remember:
Nothing obvious this week.
Improvements:
Fixed Cloudflare routing by setting up 2 tunnels vs 1.
Got the art skills in PAI up and running with the API’s. Pretty cool.
Consolidated 2 home lab pages in Notion to 1.
Created homelab diagrams with PAI art skill.
Created a video on automated update of Davinci Resolve on Arch Linux
PAI upgraded to 2.5
Summary of Today’s Linux Fixes (PAI 2.5 Migration):
Voice Server: Migrated from ElevenLabs → Qwen3-TTS (local AI, no API costs)
Updated audio_player.py: afplay → mpv --no-video
Statusline Restored (3 fixes from PAI 2.5 regressions):
Location: IP geolocation → Hardcoded
Temperature: Celsius → Fahrenheit
MCP Count: Missing → Added with 10min cache
Git Repos Migrated:
PAI/USER: 44 modified files (templates → your customizations)
MEMORY: Session learning files from today
CustomAgents: Restored with clean working tree
AI Steering Rule Added: Web server 0.0.0.0 binding + IP display
Directory Structure: CORE → PAI paths updated throughout
Wrote a skill for PAI for grabbing local files easily when I am remote over SSH by sending them to Nextcloud.
Published it to GitHub after requests for it in forums.
https://github.com/drake495/PAI-FileShare-skillUpgraded the algorithm in PAI to v0.5.3.
Created a video for installing PAI 2.5 on Arch Linux.
Updated local models for ollama. (local AI models)
qwen2.5-coder:14b │ Coding & technical │ 9.0 GB │ Best-in-class code at this tier │
phi4:14b │ Logic & math │ 9.1 GB │ Structured reasoning, underrated │
gemma3:12b │ General daily driver │ 8.1 GB │ Well-rounded, great at creative + conversation │
llama3.1:8b │ Fast fallback │ 4.9 GB │ Quick replies, leaves VRAM headroom │
All fit cleanly in 12GB.
Got Picoclaw installed in a vm and pointed to Ollama for local LLM brain to save money.
Upgraded PAI to 3.0 and 1.5 Algorithm. https://github.com/danielmiessler/Personal_AI_Infrastructure
Got Claude code + PAI running on local model for brain vs OPUS
Mistakes:
Mostly grammar and typos when publishing content. lol.
That’s it for this week’s Tech Update. -Derek
Check out my Social Media Links, my Book, my Podcast, and my Substack below: