So the model is one part of an agent. The rest has a name.
Harness
Also called Agent harness, Agentic harness
Everything in an agent except the model: the loop, the tools, the instructions, the permissions, the hooks, the sandbox. It decides what the model sees and what happens to what it asks for.
You can't change how the model thinks, but you can change nearly everything it sees and can do. The same model in a different harness does different work, and the harness is the only part of that you own.
Say it to your agent
“Don't just fix this. Tell me which part of our setup let it through: an instruction that was missing, a check that doesn't exist, or a permission that is too wide. Then propose one change to that part.”