3 ms·
But that‘s not an issue with the coding agent. It’s the model that doesn’t follow the instructions. Given how an LLM works, you can never be sure it will alway
by bhaak 6mo ago
But that‘s not an issue with the coding agent. It’s the model that doesn’t follow the instructions.
Given how an LLM works, you can never be sure it will always work. LLMs are not deterministic.
- DeathArrow 6mo agoIsn't a harness supposed to guide and steer yhe coding agent?
- bhaak 6mo agoWhile the harness can block certain actions (e.g., tool usage), it can’t enforce perfect adherence to instructions because the model itself is probabilistic. The harness can reduce deviations, but it can’t eliminate the fundamental unpredictability of LLMs. The rules that are fed into the AI are not unbreakable laws to the AI. We should always remember that.