Parallel multi-topic research with independent agents generating structured knowledge bases from any input material.

• Launches dedicated research agents for each topic in parallel, each conducting 4–6 rounds of deep retrieval with deduplication checks to avoid keyword overlap
• Accepts three input modes: file-based (/multi-search @path.md), direct paste, or explicit topic specification with custom research directions
• Generates structured output with a research overview document, individual topic reports with embedded source links, and actionable recommendations organized by priority
• Enforces strict quality standards: single-problem focus per agent, inline citation linking, narrative prose formatting, and no secondary subdirectories or redundant summary files