Implements academic research strategies and decomposition techniques to scale AI-driven investigations from simple facts to complex multi-faceted queries.
This skill provides a structured framework for Claude to perform high-quality research by applying advanced decomposition strategies such as Self-Ask, Parallel, and DAG-based planning. It ensures research effort scales proportionally with query complexity, prevents common research anti-patterns, and utilizes dynamic re-planning to adapt to intermediate findings. By bridging the compositionality gap in LLMs, it enables Claude to execute deep technical surveys, competitive analyses, and complex problem-solving tasks with professional-grade rigor and efficiency.
主な機能
019 GitHub stars
02Advanced query decomposition using Self-Ask, Least-to-Most, and DAG-based patterns
03Anti-pattern detection to prevent over-decomposition and plan rigidity
04Multi-signal stopping criteria to ensure comprehensive coverage without token waste
05Dynamic effort scaling based on complexity levels (Simple, Moderate, Complex)
06Iterative discovery and real-time research re-planning as new information emerges
ユースケース
01Synthesizing multi-hop factual answers that require data from diverse sources
02Performing deep technical architectural comparisons between competing technologies
03Conducting comprehensive market surveys or technology landscape analyses