Research Skill (Claude) — Step 1: Local + Domain Research

To generate draft requirements from code and domain knowledge, you run a research step that uses code intelligence tools and vault search to gather local implementation evidence, then translates those findings into feature and non-functional requirement documents.

Domain research for software feature planning is the process of searching existing research documents and performing web research to capture external knowledge and prior art, which is then saved to a domain research markdown file for downstream architecture decisions.

You document local implementation details by using MCP and LSP code intelligence tools to discover how a target feature should function within your specific repository, subsequently saving the collected evidence into a local research markdown file.

Yes, this research step is specifically designed for multi-LLM cooperative pipelines where Step 1 collects local implementation evidence and domain context to hand off to a downstream Codex step for further processing.

Yes, the research process explicitly captures the user's original request verbatim to ensure the generated local and domain research markdown files align perfectly with the initial product or engineering idea.

The best way to translate prior domain findings into non-functional requirements is to aggregate external knowledge and local implementation evidence into draft requirement options, accelerating downstream design and architecture decisions.