Context for LLMs in testing: from an insurance premium calculator to a hundred-page specification

▤ Summary

This article explains how to build effective context for LLMs when using them in software testing and documentation workflows. It walks through a practical five-step approach that starts with gathering source materials such as Confluence pages, CSV files, screenshots, and Markdown notes. It then shows how to convert PDFs and images into usable text with Python tools and OCR so the model can work with richer project context. The piece also discusses different input formats, including image-based Figma or UI screenshots, and emphasizes the value of structured context for better test artifacts and requirements handling.

▦ Linked Companies

Cursor

up to $1M