Project Synapse MCP Server

How to use

You interact with the Synapse MCP Server through an MCP client. In practice, you will ingest text to enrich the knowledge graph, trigger autonomous insight generation, and query the system for synthesized insights with complete reasoning trails. Use the client to send natural language input, request topic-specific insights, and explore connections across entities to reveal hidden patterns.

Typical usage patterns include ingesting raw text to populate and expand the knowledge graph, running the insight generator to produce pattern-based hypotheses with confidence scores, querying the knowledge base with complex semantic queries, and traversing connections to uncover multi-hop relationships. You will receive insight-first responses that include evidence trails and centrality or clustering analyses to help you reason about the results.

How to install

Prerequisites you must have before installing include Python 3.10 or newer, a Neo4j database, and the uv package manager.

Step 1: Install and set up the project dependencies and environment.

cd /home/ty/Repositories/ai_workspace
git clone <repository-url> project-synapse-mcp
cd project-synapse-mcp

# Create virtual environment
uv venv --python 3.12 --seed
source .venv/bin/activate

# Install dependencies
uv add -e .

Step 2: Set up the Neo4j database.

# Install Neo4j (Ubuntu/Debian)
sudo apt update
sudo apt install neo4j

# Start Neo4j service
sudo systemctl start neo4j
sudo systemctl enable neo4j

# Set password (default user: neo4j)
sudo neo4j-admin set-initial-password synapse_password

Step 3: Download the spaCy model.

uv run python -m spacy download en_core_web_sm

Step 4: Configure the environment.

cp .env.example .env
# Edit .env with your configuration

MCP client integration example

To connect an MCP client such as Claude Desktop, you configure a local stdio MCP entry that runs the server module. The following configuration shows the required runtime command, arguments, and environment variables to launch the server as an MCP endpoint.

{
  "mcpServers": {
    "project_synapse": {
      "command": "uv",
      "args": [
        "--directory",
        "/path-to-your/project-synapse-mcp",
        "run",
        "python",
        "-m",
        "synapse_mcp.server"
      ],
      "env": {
        "NEO4J_URI": "bolt://localhost:7687",
        "NEO4J_USER": "neo4j",
        "NEO4J_PASSWORD": "<your-neo4j-password>",
        "NEO4J_DATABASE": "neo4j",
        "LOG_LEVEL": "INFO"
      }
    }
  }
}

Additional notes

The server exposes a set of core tools that enable end-to-end semantic processing, knowledge graph management, and autonomous insight generation. You can ingest text, trigger insight generation, query knowledge with natural language, explore connections, and analyze semantic structures.

ingest_text

Process and analyze text using the full semantic pipeline: ingest raw text, apply Montague Grammar parsing, extract entities, store in the knowledge graph, and autonomously generate insights.

generate_insights

Trigger autonomous insight generation with pattern detection, community analysis, semantic clustering, and confidence-scored hypotheses.

query_knowledge

Query knowledge using natural language with an insight-first focus, providing complete reasoning trails.

explore_connections

Traverse the graph to reveal multi-hop relationships and analyze connection strengths.

analyze_semantic_structure

Perform deep semantic analysis with Montague Grammar including logical form generation and entity-relationship extraction.

synapse://knowledge_stats

Real-time statistics on the knowledge graph, insights, and system health.

synapse://insights

Topic-specific insight retrieval with evidence trails and confidence scores.