search_dashboards MCP tool for pilot context workflows

Problem

AI agents working through MCP have no way to discover existing dashboards or the validated query patterns they contain. When an analyst asks an agent to build a dashboard for a topic that already has coverage, the agent starts from scratch rather than referencing existing work. This wastes effort, produces inconsistent metrics definitions, and misses the opportunity to reuse query patterns that have already been validated by human analysts.

Context

• MCP exposes search_dashboards with useful filters and ranked results.
• Results include enough metadata to reuse dashboard/query patterns quickly.
• Pilot workflows demonstrate reduced time-to-first-draft using search results.

Possible Solutions

Plan

• Define tool input/output contract and ranking defaults for pilot needs.
• Implement index/query path for dashboard corpus search.
• Add tests for ranking relevance and error handling.
• Document agent prompt patterns that incorporate search results.

Implementation Progress

• dashboard-factory published packs metadata quality

Review Feedback

• [ ] Review cleared