Archives
- 10 Jun rocket-cli: a Rocket.Chat MCP server with a local FTS5 brain
- 26 May Design Fluency Meets the Knowledge Graph
- 22 May How a Cryptominer Spent Two Days on My Server — and How I Found It
- 18 May When your VLM test flake is actually a VNC capture race
- 15 May Turbo3 + MTP: Merging Two llama.cpp Forks
- 13 May Qwen 3.6 Dense vs MOE on Local Stack: what MTP actually delivers
- 13 May Qwen 3.6 27B with Native MTP on llama.cpp
- 09 May Running Qwen 3.6 35B MoE on an RTX 3060 12GB via -ncmoe
- 08 May 1.5× Faster Agentic Coding with MTP on Qwen 3.6 27B
- 08 May issuer PR Dashboard: Web UI on Top of SQLite + LLM Reviews
- 08 May Extending Issuer to Pull Requests: Same Pattern, Different Data
- 07 May Running Issuer at Scale: Closing 42 Issues Without Losing Control
- 07 May issuer: A Local SQLite-Backed GitHub Issue Manager Powered by LLMs
- 06 May When Ghostty starts spitting escape codes at every keypress
- 04 May Club 3090 vs My Llama Setup
- 02 May Stopping a VLM-Driven Test From Leaking My Password
- 29 Apr Running NVIDIA Nemotron 30B with Vision on a 24 GB GPU
- 29 Apr Nemotron 30B on 24 GB: Benchmarks and a Quantization Quirk
- 29 Apr Qwen3.6-27B on an M1 Max: when the laptop config matches the 3090
- 28 Apr Taming Qwen Overthinking with GBNF Grammars
- 28 Apr RTX 3090 Power Limit: Finding the Sweet Spot for Local LLM Inference
- 28 Apr Bootstrapping this blog, and the skill that writes it