corpus-investigation

What problem does it solve?

This Skill enables systematic, reproducible investigation of very large document corpora (100GB+) using a token-efficient methodology. It outputs structured section analyses, metadata schemas, and RAG guidance to help design scalable knowledge bases without exhaustive reading.

Core Features & Use Cases

• 5-phase investigation framework (Reconnaissance, Stratified Sampling, Pattern Verification, Edge Case Analysis, Synthesis) to produce a complete Section Analysis Document.
• Stratified sampling across size, time, type, and depth to minimize token usage while preserving representative patterns.
• Computational pattern verification using grep, find, and other shell tools to quantify coverage across thousands of files without exhaustive reading.
• Output-ready Section Analysis Document with 5-layer metadata schema and RAG integration recommendations for teams building large-scale knowledge bases.

Quick Start

To begin, say: "Investigate the corpus at /path/to/corpus/". Claude will activate this skill and return a comprehensive Section Analysis Document with reproducible commands and patterns.