George
@georgesl.bsky.social
Centminmod.com LEMP stack maintainer. Cloudflare Community MVP since 2018
Pinned
Running Claude Code + OpenAI Codex CLI + Google Gemini CLI + OpenCode in a Microsoft Visual Studio Code dev container running a Debian 12 Docker image. Created my VSC dev container guide claude-devcontainers.centminmod.com 🤓 Enjoy and repost 😎
November 11, 2025 at 10:37 PM
Running Claude Code + OpenAI Codex CLI + Google Gemini CLI + OpenCode in a Microsoft Visual Studio Code dev container running a Debian 12 Docker image. Created my VSC dev container guide claude-devcontainers.centminmod.com 🤓 Enjoy and repost 😎
Love AI and all, but it's impact on Nand flash and memory and storage prices is crazy. A few months ago bought 2x 2TB Kingston KC3000 NVMe drives as spares for AUD$205/209 each. Now you can't even get 1TB KC3000 NVMe for less than AUD$237 😲
November 11, 2025 at 2:49 PM
Love AI and all, but it's impact on Nand flash and memory and storage prices is crazy. A few months ago bought 2x 2TB Kingston KC3000 NVMe drives as spares for AUD$205/209 each. Now you can't even get 1TB KC3000 NVMe for less than AUD$237 😲
WTH, didn't know GitHub has a repo watch limit!
November 10, 2025 at 10:26 PM
WTH, didn't know GitHub has a repo watch limit!
Claude Code definitely needs >200K token context window sizes
November 10, 2025 at 5:24 PM
Claude Code definitely needs >200K token context window sizes
GlassWorm Malware Discovered in Three VS Code Extensions with Thousands of Installs. Uses invisible Unicode characters to hide malicious code in code editors and abuses the pilfered credentials to compromise additional extensions and further extend its reach thehackernews.com/2025/11/glas...
GlassWorm Malware Discovered in Three VS Code Extensions with Thousands of Installs
GlassWorm malware returns in VS Code extensions, hiding via Unicode and reviving itself through blockchain.
thehackernews.com
November 10, 2025 at 4:05 PM
GlassWorm Malware Discovered in Three VS Code Extensions with Thousands of Installs. Uses invisible Unicode characters to hide malicious code in code editors and abuses the pilfered credentials to compromise additional extensions and further extend its reach thehackernews.com/2025/11/glas...
Moonshot Kimi K2 with new agent mode early beta
November 9, 2025 at 9:07 PM
Moonshot Kimi K2 with new agent mode early beta
For folks still interested in local AI inference with 2x AMD Ryzen AI Max Strix Halo mini pc with 2x128GB = 256GB memory combined
youtu.be/0cIcth224hk?...
youtu.be/0cIcth224hk?...
Building a Two-Node AMD Strix Halo Cluster for LLMs with llama.cpp RPC (MiniMax-M2 & GLM 4.6)
YouTube video by Donato Capitella
youtu.be
November 9, 2025 at 8:36 PM
For folks still interested in local AI inference with 2x AMD Ryzen AI Max Strix Halo mini pc with 2x128GB = 256GB memory combined
youtu.be/0cIcth224hk?...
youtu.be/0cIcth224hk?...
Trying Claude Code for the web and definitely slower than local Claude Code CLI/VSC extensions with lack of MCP support right now. Web version can't utilise Claude Code's OpenTelemetry usage metrics. Claude Code CLI within VSC dev container track my Claude Code token and cost usage in Grafana 😎
November 9, 2025 at 4:06 PM
Trying Claude Code for the web and definitely slower than local Claude Code CLI/VSC extensions with lack of MCP support right now. Web version can't utilise Claude Code's OpenTelemetry usage metrics. Claude Code CLI within VSC dev container track my Claude Code token and cost usage in Grafana 😎
Anyone experiencing very slow performance for Kimi K2 Thinking model via Openrouter AI provider in Kilo Code? Asked it to improve my 8000 lines long VSC dev container HTML guide, it's at a snail slow pace only 1/3 of way through 1st draft vs Claude Code Sonnet 4.5 is already completing 7th draft!
November 9, 2025 at 3:24 PM
Anyone experiencing very slow performance for Kimi K2 Thinking model via Openrouter AI provider in Kilo Code? Asked it to improve my 8000 lines long VSC dev container HTML guide, it's at a snail slow pace only 1/3 of way through 1st draft vs Claude Code Sonnet 4.5 is already completing 7th draft!
My Debian 12 Docker based Microsoft Visual Studio Code dev container on macOS where these tools can party together in an isolated container - OpenAI Codex CLI, Gemini CLI, Claude Code, OpenCode CLI, Vercel CLI, Cloudflare Wrangler, AWS CLI 🤓
November 9, 2025 at 12:03 PM
My Debian 12 Docker based Microsoft Visual Studio Code dev container on macOS where these tools can party together in an isolated container - OpenAI Codex CLI, Gemini CLI, Claude Code, OpenCode CLI, Vercel CLI, Cloudflare Wrangler, AWS CLI 🤓
Added OpenCode CLI and Visual Studio Code extension to my Debian 12 Docker based VSC dev container so that all these tools can party together in an isolated container - OpenAI Codex CLI, Gemini CLI, Claude Code, Vercel CLI, Cloudflare Wrangler, AWS CLI 🤓 Claude Code session start up detected info.
November 9, 2025 at 1:53 AM
Added OpenCode CLI and Visual Studio Code extension to my Debian 12 Docker based VSC dev container so that all these tools can party together in an isolated container - OpenAI Codex CLI, Gemini CLI, Claude Code, Vercel CLI, Cloudflare Wrangler, AWS CLI 🤓 Claude Code session start up detected info.
Theo's take on Moonshot Kimi K2 Thinking LLM model www.youtube.com/watch?v=ZP9l...
Kimi K2 is the best model ever (kind of…)
YouTube video by Theo - t3․gg
www.youtube.com
November 8, 2025 at 3:25 PM
Theo's take on Moonshot Kimi K2 Thinking LLM model www.youtube.com/watch?v=ZP9l...
My Claude Code starter template Github repo with CLAUDE.md memory bank system, hooks, subagents, slash commands, skills just broke 1300+ Github stars and still consistently gets 800-1000 unique visitors/week according to Github github.com/centminmod/m... 🤓
November 8, 2025 at 2:04 PM
My Claude Code starter template Github repo with CLAUDE.md memory bank system, hooks, subagents, slash commands, skills just broke 1300+ Github stars and still consistently gets 800-1000 unique visitors/week according to Github github.com/centminmod/m... 🤓
This is Claude Code + Openai Codex CLI + Gemini CLI in one Visual Studio Code multi-tab supported dev container with full visual IDE/MCP support 🤓 Best of both worlds CLI + GUI editor 😎
November 8, 2025 at 1:50 PM
This is Claude Code + Openai Codex CLI + Gemini CLI in one Visual Studio Code multi-tab supported dev container with full visual IDE/MCP support 🤓 Best of both worlds CLI + GUI editor 😎
Some folks misunderstand that AI CLI clients are just CLI tools and don't realise you can pair them with a GUI IDE. This is Claude Code + Openai Codex CLI + Gemini CLI in one Visual Studio Code multi-tab supported dev container full visual IDE/MCP support 🤓 Best of both worlds CLI + GUI editor 😎
November 8, 2025 at 1:07 PM
Some folks misunderstand that AI CLI clients are just CLI tools and don't realise you can pair them with a GUI IDE. This is Claude Code + Openai Codex CLI + Gemini CLI in one Visual Studio Code multi-tab supported dev container full visual IDE/MCP support 🤓 Best of both worlds CLI + GUI editor 😎
Trying out new OpenAI gpt-5-codex-mini in Codex CLI 0.56 in Visual Studio Code dev container running Debian 12 🤓
November 7, 2025 at 7:11 PM
Trying out new OpenAI gpt-5-codex-mini in Codex CLI 0.56 in Visual Studio Code dev container running Debian 12 🤓
So I learnt from Claude Code that session start hook output is wrapped in a <system-reminder> tag with the prefix: SessionStart:startup hook success: [output here]. That's why it only shows output if you use claude CLI with --verbose flag. 🤓
November 7, 2025 at 5:30 PM
So I learnt from Claude Code that session start hook output is wrapped in a <system-reminder> tag with the prefix: SessionStart:startup hook success: [output here]. That's why it only shows output if you use claude CLI with --verbose flag. 🤓
Google debuts Ironwood 7th generation TPU AI chips with 4X performance boost - training and inference workloads compared to its predecessor. 🤓 venturebeat.com/ai/google-de...
venturebeat.com
November 7, 2025 at 1:25 PM
Google debuts Ironwood 7th generation TPU AI chips with 4X performance boost - training and inference workloads compared to its predecessor. 🤓 venturebeat.com/ai/google-de...
Moonshot Kimi K2 Thinking AI LLM model www.youtube.com/watch?v=5t3k...
Kimi K2 Thinking Is The BEST Open Source Model - First Look & Testing
YouTube video by Bijan Bowen
www.youtube.com
November 6, 2025 at 10:16 PM
Moonshot Kimi K2 Thinking AI LLM model www.youtube.com/watch?v=5t3k...
Apple gave up on in-house AI, paying Google for a custom 1.2 trillion parameter LLM model for Apple Siri www.youtube.com/watch?v=LwPN... 😲
Apple is paying Google to fix Siri (yes really)
YouTube video by Theo - t3․gg
www.youtube.com
November 6, 2025 at 12:13 PM
Apple gave up on in-house AI, paying Google for a custom 1.2 trillion parameter LLM model for Apple Siri www.youtube.com/watch?v=LwPN... 😲
Claude Code + Codex CLI + Gemini CLI Visual Studio Code dev container with Claude Code session start hook to inform user whether their chat session is in a dev container, etc 🤓😎
November 5, 2025 at 7:11 PM
Claude Code + Codex CLI + Gemini CLI Visual Studio Code dev container with Claude Code session start hook to inform user whether their chat session is in a dev container, etc 🤓😎
Just remember Claude Code for Web bug report and feedback can use your chat and data for model training purposes. So maybe Github repo issue report if you have concerns? privacy.claude.com/en/articles/...
November 5, 2025 at 9:18 AM
Just remember Claude Code for Web bug report and feedback can use your chat and data for model training purposes. So maybe Github repo issue report if you have concerns? privacy.claude.com/en/articles/...
Claude AI Pro and Max users getting FREE usage credits for Claude Code on the web and iOS app 😎👍
Available for a limited time (til Nov 18):
- Max users: $1,000 in credits
- Pro users: $250 in credits
Details: support.claude.com/en/articles/...
Available for a limited time (til Nov 18):
- Max users: $1,000 in credits
- Pro users: $250 in credits
Details: support.claude.com/en/articles/...
Claude Code Promotion | Claude Help Center
support.claude.com
November 4, 2025 at 5:49 PM
Claude AI Pro and Max users getting FREE usage credits for Claude Code on the web and iOS app 😎👍
Available for a limited time (til Nov 18):
- Max users: $1,000 in credits
- Pro users: $250 in credits
Details: support.claude.com/en/articles/...
Available for a limited time (til Nov 18):
- Max users: $1,000 in credits
- Pro users: $250 in credits
Details: support.claude.com/en/articles/...
Vibe coding in isolated Microsoft Visual Studio dev containers with Claude Code + OpenAI Codex CLI + Google Gemini CLI and Cloudflare Wrangler, Vercel CLI, AWS CLI, all in a self-contained Docker container 🤓 Though each dev container can use up to 3.9GB of memory.
November 4, 2025 at 2:51 PM
Vibe coding in isolated Microsoft Visual Studio dev containers with Claude Code + OpenAI Codex CLI + Google Gemini CLI and Cloudflare Wrangler, Vercel CLI, AWS CLI, all in a self-contained Docker container 🤓 Though each dev container can use up to 3.9GB of memory.
I missed that context7 MCP server and documentation prefer setting an API key now. My existing context7 MCP servers seem to fail to connect sometimes without it now? Seems change in docs was at least 3 months ago github.com/upstash/cont...
docs: add API key support for higher rate limits and authentication · upstash/context7@6202ad8
github.com
November 4, 2025 at 1:42 PM
I missed that context7 MCP server and documentation prefer setting an API key now. My existing context7 MCP servers seem to fail to connect sometimes without it now? Seems change in docs was at least 3 months ago github.com/upstash/cont...