Thanks for laying out the timeline so clearly. I’ve been using Ollama for months because it ‘just worked,’ but I had no idea about the missing attribution or the hashed blob storage lock-in. The fact that I can’t easily point another tool at the same GGUF file is a dealbreaker for me. Going to try llama-server today.
Really impressive results. The point about the harness mattering more than the model is spot on — we've seen similar patterns in our own work.
One thing that stood out to me is your use of hash-anchored edits + AST-based context selection. We're building something in a similar direction with the Sovereign AI Stack, but with a stronger focus on governance and verification.
Curious — did you run into issues with context drift when using AST queries on very large codebases? We found that combining it with incremental symbol DB updates helped a lot.