yellow-research

Inline code research for active development. Use when user asks how a library works, needs code examples, API patterns, or framework documentation. Returns synthesized answer in-context — no file saved.

Multi-source deep research saved to docs/research/. Use when user needs a comprehensive report, competitive analysis, technical landscape overview, or architectural decision support. Saves output as docs/research/<slug>.md.

Check which research API keys and MCP sources are configured and active. Use when first installing the plugin, after adding API keys, or to understand why a research command degraded to fewer sources.

Enrich an existing plan with codebase validation and external research, annotating inline. Use when a plan needs deeper validation before starting /workflows:work.

Inline code research for active development. Use when user asks how to use a library, needs code examples, API patterns, or framework documentation. Routes to best source by query type; returns concise in-context synthesis without saving a file.

Routes /research:deep queries across multiple MCP sources. Use when deep research needs multi-source investigation. Triages complexity and dispatches parallel fan-out -- simple queries go to Perplexity alone; moderate to 2-3 parallel sources; complex topics trigger full fan-out including Parallel Task MCP for async reports.

Reference conventions for yellow-research plugin — slug naming, output format, source selection, API key setup, graceful degradation, and when to compound findings.