Trust the model: A minimalist approach to using LLMs for coding with clear specs
By
yehiaabdelm
5h ago· 3 min readenOpinion
Summary
The author argues that the best approach to using LLMs for coding is to minimize external orchestration and instead trust the model's capabilities. The key is providing clear, well-structured specifications that focus on intent, input/output contracts, constraints, and preconditions — using the AskUserQuestion tool for disambiguation. The author draws an analogy: like a less intelligent person giving instructions, brevity and clarity of purpose are more valuable than verbose step-by-step directions.
Source
Key quotes
· 5 pulledI think the value right now is to focus less on external orchestration if at all. trust the (current best) model to do it better than anything you bolt on to the harness.
focus your energy on providing clearer specs.
I think the optimal spec is a disambiguated (through liberal use of the AskUserQuestion tool) 1 intent, 2, input/output contracts 3 constraints and 4 preconditions.
focus on that and get out of the models way.
imagine a person who was not as smart as you was trying to tell you how to do a task. would you want more verbosity and step by step instructions or would you want them to just cut to the chase
I think the value right now is to focus less on external orchestration if at all. trust the (current best) model to do it better than anything you bolt on to the harness. focus your energy on providing clearer specs. I think the optimal spec is a disam

Comments
Sign in to join the conversation.
No comments yet. Be the first.