Overview

The Atlassian Remote MCP server's search and fetch tools currently return lower-quality semantic results than the Rovo interface in product, particularly when queries exceed 10-15 words. Users are seeing a significant drop in relevance for complex queries, suggesting a lack of robust post-processing or re-ranking of results.

Current Process

The current MCP server implementation utilizes a search tool that performs well for very short, concise queries but fails to maintain accuracy or provide sophisticated semantic understanding for longer, more detailed search strings.

Feature Request

We would like the Atlassian Remote MCP server to be updated with the same robust semantic search, post-processing, and result re-ranking capabilities found in the native Rovo experience to ensure high-relevance results for complex workloads.