I think that is what OP was fishing for. Wrong answers ;).
Llama.cpp is not harness, setting up prompt template to have instruct model is also not what people call harness.
What people call harness nowadays is software that you interact with and it proxies model interactions. It can do agentic loop but the loop is not really much use for models that do “thinking” unless you want to layer your own loop on top for dealing with results and good harness can help - but usually if something runs loop it is called an agent not a harness.
So if harness proxies all interaction with model it also provides tools for model. It can provide context management and security features, where as security features would be preventing injections and making sure model doesn’t access stuff it shouldn’t.
None of these categories seem to build any meaningful rapport. Any honest answer I give is deeply unsatisfying to both parties, and we typically have better data from telemetry or crash logs than the emails provide. It’s certainly useful for us to receive them, but there’s not a helpful response I can give.
I don't know how you read the blog post but it seems your interpretation is uncharitable to say the least.
He wrote he already has good enough telemetry, there is just not much he can write in an e-mail to customer, besides generic "we know the issue thank you for the report" that would be useful for the customer and for themselves.
My favorite example is how NATO recognized cyberspace as operational domain in 2016 — like no one there watched or heard about a movie from 1983 called „War Games”.
I think you vastly overestimate speed with which people acknowledge new facts.
Second best is only Catholic Church fully acknowledging heliocentric theories in 1992.
Well not really. I couldn’t care less. I would move out anyway.
I just have seen firsthand how people who claim to be „for the local community” on the posters when it is election time — doing exactly the opposite otherwise.