Synthèse
And sometimes the agent didn’t need another tool at all. It needed instruction. Prompts started to make more sense here, less as magic or a way to make the model smarter, but as workflow affordances. If a human analyst knows to triage first, inspect bounded evidence second, retrieve remediation guidance third, and only export bulk data when necessary, we should not expect the agent to rediscover that ordering every time. The best prompts are not stringing a complicated sequence of steps together
And sometimes the agent didn’t need another tool at all. It needed instruction. Prompts started to make more sense here, less as magic or a way to make the model smarter, but as workflow affordances. If a human analyst knows to triage first, inspect bounded evidence second, retrieve remediation guidance third, and only export bulk data when necessary, we should not expect the agent to rediscover that ordering every time. The best prompts are not stringing a complicated sequence of steps together (see the export URL above).
The prompts are to explain, guide and direct attention to the parts of the application that matter in the context of the problem the user is trying to solve.