research-loom-narrow

Files (2)

• SKILL.md
• openai.yaml · agents/

SKILL.md (first lines)

# Topic Narrowing

Use the interview brief. For each goal, identify one phenomenon or relationship, a bounded context/population/system, and the outcome or knowledge sought. Identify candidate interpretations and ask the user to resolve consequential ambiguity.

Propose a small number of focused alternatives when needed, explaining the tradeoff in scope, evidence availability, and feasibility. Challenge questions that bundle distinct outcomes, assume a desired answer, or require unavailable evidence. Preserve the user's goal; do not change an empirical paper to a review without their agreement.

Produce one precise question per goal with scope, exclusions, key concepts, candidate synonyms, evidence needed, and a completion criterion. Show the chosen wording and get the user's confirmation before substantive searches, unless that question was already agreed in the session. If multiple goals remain, keep separate question records and suggest a distinct search plan for each.

From amaljithkuttamath/research-loom, MIT.

More from amaljithkuttamath/research-loom