📝 MCP Server Guide

This page is the MCP hub. Detailed instructions are split into dedicated pages to avoid duplicate setup content.

🚀 What MCP gives you

  • Semantic search in your ingested knowledge base
  • Hierarchy-aware retrieval for structured docs
  • Attachment-focused search
  • Integration with Cursor, Windsurf, Claude Desktop, and other MCP clients

⚙️ Client configuration links

🎯 Prerequisites

  • Ingestion completed at least once with qdrant-loader ingest
  • QDrant reachable from your MCP runtime
  • LLM provider configured

Configuration references:

⚡ Quick run

mcp-qdrant-loader

For production transport and worker tuning, use Setup & Integration Guide.

🔍 Multi-Tool Search Strategies

Complete feature investigation

  1. Start with Semantic Search to understand the topic.
  2. Use Hierarchy Search to explore document structure.
  3. Apply Relationship Analysis to map dependencies.
  4. Use Conflict Detection to identify inconsistencies.

Documentation quality audit

  1. Use Hierarchy Search for structure and gap analysis.
  2. Use Conflict Detection for inconsistency checks.
  3. Use Similarity Detection to review duplication.
  4. Use Complementary Content to assess completeness.

Implementation planning

  1. Use Semantic Search for patterns and examples.
  2. Use Complementary Content for supporting references.
  3. Use Relationship Analysis for dependency understanding.
  4. Use Clustering to organize related materials.

🚀 Performance Optimization

Search efficiency

  • Use specific queries instead of broad terms.
  • Apply source/type filters when appropriate.
  • Use practical limits for cross-document analysis.

Result quality

  • Provide context in your query.
  • Prefer natural language for semantic retrieval.
  • Combine tools to improve coverage and precision.

🔍 Quick validation

In Cursor/Claude/Windsurf, ask a simple query like:

"Find setup notes for QDrant Loader in my ingested docs"

If the tool returns results from your indexed content, MCP integration is working.

🧪 Integration Checklist

Setup requirements

  • QDrant Loader installed and configured
  • Documents ingested into QDrant
  • MCP server package installed
  • AI tool with MCP support (Cursor/Windsurf/Claude)
  • LLM API key configured

Configuration

  • MCP server added to client config
  • Environment variables set correctly
  • Collection name matches ingested content
  • Connection verified from AI tool

Functionality testing

  • Basic semantic search works
  • Hierarchy search navigates structure
  • Attachment search returns expected files
  • Cross-document analysis returns relationships
  • Performance is acceptable for daily usage

Team deployment

  • Configuration standardized across team
  • Best practices documented and shared
  • Security considerations reviewed
  • Troubleshooting procedures documented

🔧 Troubleshooting paths