Graphistry

You can run a GPU-accelerated graph visualization and analytics MCP server that integrates Graphistry with the MCP framework. This server lets you visualize and analyze graph data from your LLM workflows using a single, LLM-friendly graph_data API, with credentialed access to Graphistry’s visualization features.

How to use

Launch the MCP server using the standard MCP client workflow. You will connect to the server using a local, stdio-based process that is started through a package command. The server accepts a single graph_data payload to visualize graphs or hypergraphs and then performs layout, pattern detection, and advanced analytics through Graphistry.

How to install

Prerequisites you need to have installed before starting: Node.js and npm (for running the MCP client and the package), Python (for the underlying MCP runtime if you use the Python path), and an active Graphistry account for visualization.

Install and run the MCP server via the quick start path using npm to fetch and execute the package without a global install.

# Quick start: run the Graphistry MCP server without a global install
npx -y @silkspace/graphistry-mcp

# Or install globally and run
npm install -g @silkspace/graphistry-mcp
graphistry-mcp

Additional setup and configuration

Sign up for a free Graphistry account and set your credentials in your environment or in a local .env file before starting the server.

Create and edit the MCP configuration to point to your environment and credentials. The server expects Graphistry credentials and can operate in stdio mode or HTTP mode depending on how you configure the MCP client.

Security and credential handling

The server loads credentials from environment variables or a .env file using a Python-based loader, so you can securely provide credentials for local development.

Examples: visualizing a graph with the LLM-friendly API

You pass a single graph_data dictionary to the main tool visualize_graph to render a graph or hypergraph and apply layouts or analyses.

Available tools

visualize_graph

Render a graph or hypergraph using Graphistry's GPU-accelerated visualization, supporting standard graph data and hypergraphs.

get_graph_ids

List all stored graph IDs for the current session.

get_graph_info

Retrieve metadata for a stored graph, including node/edge counts, title, and description.

apply_layout

Apply a layout to a graph, such as force_directed, radial, circle, or grid.

detect_patterns

Run network analytics like centrality, community detection, path finding, and anomaly detection.

encode_point_color

Encode node color by a specified column (categorical or continuous).

encode_point_size

Encode node size by a specified column (categorical or continuous).

encode_point_icon

Encode node icons by a column with optional mappings.

encode_point_badge

Encode node badges by a column with optional mappings.

apply_ring_categorical_layout

Arrange nodes in rings by a categorical column.

apply_group_in_a_box_layout

Arrange nodes in a group-in-a-box layout (requires igraph).

apply_modularity_weighted_layout

Arrange nodes by modularity-weighted layout (requires igraph).

apply_ring_continuous_layout

Arrange nodes in rings by a continuous column.

apply_time_ring_layout

Arrange nodes in rings by a datetime column.

apply_tree_layout

Arrange nodes in a tree layout (layered hierarchical).

set_graph_settings

Adjust advanced visualization settings such as point size and edge influence.