Deep Research

What it does

A structured 7-phase research protocol that turns high-level questions into decision-grade, auditable reports. The system classifies requests, scopes the problem, forms hypotheses, executes targeted multi-source searches, triangulates claims, synthesizes findings, runs QA checks, and packages outputs with full source catalogs and evidence ledgers. Outputs are designed for executive or technical consumption and can be delivered as a report, data pack, or machine-readable artifacts.

When to use it

Use Deep Research when you need more than a quick fact: multi-source synthesis, judgment-based analysis, or complex investigation where claims must be verifiable. Typical triggers: deep research [topic], research [topic] thoroughly, investigate [topic]. Not suitable for single-fact lookups.

What's included

• Scripts: none detected in repo (has_scripts=false)
• References: none bundled (has_references=false)
• Instructions: detailed 7-phase protocol (Classification → Scoping → Hypothesis → Plan → Query → Triangulation → Synthesis → QA → Packaging). Includes guidance on query planning, source scoring, 2-source verification for critical claims, output folder layout, and QA checklist.

Compatible agents

Best suited for agent systems that support multi-step orchestration, citation tracking, and long-context workflows (e.g., Claude/Claude Code, multi-agent orchestrators, or custom pipeline runners).