home / mcp / expert registry mcp server

Expert Registry MCP Server

High-performance MCP server for expert discovery with vector and graph database integration. Features Docker containerization, multi-client architecture, and comprehensive expert management tools.

Installation

Add the following to your MCP client configuration file.

Configuration

View docs

{
  "mcpServers": {
    "agentience-expert-registry-mcp": {
      "command": "fastmcp",
      "args": [
        "run",
        "expert-registry-mcp"
      ],
      "env": {
        "NEO4J_URI": "bolt://localhost:7687",
        "NEO4J_PASSWORD": "password",
        "EXPERT_SYSTEM_PATH": "PATH_TO_EXPERT_SYSTEM"
      }
    }
  }
}

You run a specialized MCP server that enables expert discovery, registration, and context enhancement. It integrates fast retrieval with vector and graph databases to help you find the right experts, keep contextual knowledge fresh, and form effective teams.

How to use

Start the server and connect your MCP client to access registry data, semantic search, and context injection features. You can run the server from a local development setup or deploy it as a container. Use the provided commands to start a local instance or connect to a remote MCP endpoint. The server exposes multiple tools for registering experts, performing hybrid searches that combine vector similarity with graph connectivity, loading expert contexts, injecting context into prompts, and tracking analytics.

How to install

Prerequisites: you need Python 3 and a working environment to install and run Python packages. You can also run the server in a container when you prefer production-grade deployment.

Install and run locally using Python and the FastMCP workflow:

# Create a virtual environment and install
uv venv
uv pip install -e .

# Or install directly from the package index
uv pip install expert-registry-mcp

# Start the server using the FastMCP CLI
fastmcp run expert-registry-mcp

# Or start via Python module
python -m expert_registry_mcp.server

Additional setup and dependencies

Configure environment variables for local development to connect to the vector and graph databases and to point to your expert system data.

export EXPERT_SYSTEM_PATH=/path/to/expert-system
export NEO4J_URI=bolt://localhost:7687
export NEO4J_PASSWORD=password

Available tools

expert_registry_list

List experts with filtering options to narrow results.

expert_registry_get

Retrieve detailed information for a specific expert by ID.

expert_registry_search

Search experts by query across metadata, domains, and capabilities.

expert_detect_technologies

Detect project technologies to inform expert matching.

expert_select_optimal

Select the best expert for a task using scoring heuristics.

expert_assess_capability

Assess an expert's capability against specified requirements.

expert_smart_discover

AI-powered hybrid search combining vector similarity and graph connectivity.

expert_semantic_search

Search for experts using natural language understanding.

expert_find_similar

Find similar experts based on embeddings.

expert_explore_network

Explore expert relationships within the graph.

expert_find_combinations

Find complementary expert teams that meet requirements.

expert_load_context

Load expert knowledge context for prompt augmentation.

expert_inject_context

Enhance prompts with expertise injected at specific points.

expert_track_usage

Record expert performance and task outcomes.

expert_get_analytics

Retrieve analytics and performance metrics for an expert.