fusion-research

Fusion research classifies questions into Framework, EDS, Skills catalog, or Docs domains and routes them to the appropriate local agent. It uses MCP-backed retrieval to return metadata source paths and supporting excerpts for API definitions, component props, and governance statements.

Fusion research requires MCP-backed retrieval tools including mcp_fusion_search_framework, mcp_fusion_search_eds, mcp_fusion_search_skills, and mcp_fusion_search_docs. Documented fallbacks are also supported when these primary MCP retrieval endpoints are unavailable.

You can verify Fusion platform onboarding steps and skill scope boundaries by orchestrating source-backed research across the skills catalog and platform documentation. The process mandates capturing metadata source and supporting excerpts to provide factual validation with provenance.

Fusion research removes uncertainty by orchestrating source-backed lookups across Framework APIs, EDS components, the skills catalog, and platform documentation. It mandates captured metadata source and supporting excerpts with a single refinement pass, clearly stating assumptions and uncertainty rather than guessing.

Domain classification automatically categorizes queries into Framework, EDS, Skills catalog, or Docs domains and routes them to the appropriate local agent. This ensures MCP-backed retrieval targets the correct source for returning evidence-backed answers with captured metadata.

When researching EDS component props, expect stated assumptions and uncertainty if the MCP-backed retrieval yields incomplete evidence. The research process enforces a single refinement pass and clearly states these assumptions rather than returning unsupported guesses.