Deep Research Agent Brief

Research this thoroughly: [THE QUESTION]

Why I'm asking + the decision it feeds: [THE CONTEXT: this sets depth and angle]
Scope boundaries: [TIMEFRAME THAT MATTERS (e.g. "2024-present"), GEOGRAPHY, WHAT'S EXPLICITLY OUT OF SCOPE]
Output: [REPORT LENGTH + FORMAT, e.g. "2-page brief with a comparison table"]

Before searching, do two things:
1. **Decompose**: break the question into 3-6 sub-questions that together answer it. Show me the list. If my question is ambiguous in a way that changes the research (which market? which definition of X?), ask now: one clarifying round beats a report about the wrong thing.
2. **Plan the search**: for each sub-question: the search angle, and what kind of source would settle it.

Research rules:
- **Source hierarchy, enforced:** primary sources (official docs, standards bodies, original studies, company filings) > reputable secondary reporting > everything else. SEO listicles and vendor marketing pages are leads to primary sources, never citations themselves. Note the publication date on every source; flag anything stale for the timeframe I set.
- **Broad, then narrow:** start with short broad queries to map the territory, then progressively specific ones. Vary search phrasings: one phrasing samples one slice of the web.
- **Citation discipline:** every non-obvious claim in the report carries its source, and the source must actually say the thing, never attach a citation to a sentence it doesn't support. Distinguish plainly: [ESTABLISHED: multiple quality sources agree] / [SINGLE-SOURCE: one source, graded] / [INFERRED: my synthesis, reasoning shown].
- **Contradictions are findings:** when sources disagree, present both with dates and authority, and explain the likely reason (methodology, timeframe, incentive), never silently pick one.
- **Stop rule: define "enough":** a sub-question is DONE when [N, default 2-3] independent quality sources agree, or when further searches return the same results. When done, stop. Do not keep searching for comfort. Cap total effort at roughly [BUDGET, e.g. "20 searches"]; if the budget exhausts with sub-questions open, report them as open rather than padding.

Report structure:
1. Bottom line (the answer in 3-4 sentences, confidence stated).
2. Findings per sub-question, each with its evidence grade and citations.
3. Contradictions and uncertainties: what's genuinely unsettled.
4. What I'd research next with more budget, ranked.
5. Source list with one-line quality notes.