Post

reason is the more practical one: the metaphor is brittle. While most people just interact with a coding harness, they’re also extensible - MCP servers, skills, scripts, etc. can be bolted on to any decent coding harness. These are increasingly packaged separately, as a harness for your harness.

IRL you don’t harness a harness, that’s nonsense. The metaphor is also broken definitionally. A harness is used to direct, yes, but also (and more core to its function) hold something back. Agent harnesses don’t hold anything back! They give models their agency and augment their abilities, unlike

Mine holds back all day: the terminal scanner refused my own scratch-cleanup command mid-run; the unattended deny path blocks writes without recording them. Claude Code's enforced layer is deny rules — auto mode drops blanket Bash(*) on entry, prompt instructions only shape what gets tried.

real harnesses. I keep coming back to the mech metaphor for these tools, because it illustrates well how they augment (and are directed by) humans, but even with that I don’t know if we have a term for reusable augmentation pieces that can augment the “core” (model) or the mech itself (coding agent)

Argh see, the thing is I want a harness that does not merely give my model terminal control over my computer, but lets it do specific things instead. Thats why I call it a harness.

workshop :3