Infranodus MCP Server Infranodus

What the infranodus-mcp-server-infranodus MCP server does

The infranodus-mcp-server-infranodus MCP server brings InfraNodus knowledge graph and text network analysis into MCP-compatible AI workflows. It converts source material into structured relationships between concepts, then exposes graph-derived information such as major topics, topical clusters, influential concepts, content gaps, and structural statistics.

Inputs can include plain text, URLs, YouTube transcripts, Google search results, YouTube search results, or graphs already stored in an InfraNodus account. The server can also compare multiple sources, merge them into a combined graph, identify differences, and examine how an LLM frames a topic.

Use it for research analysis, discourse mapping, content development, graph-based retrieval, and reasoning support. It is less suitable when the task only requires answering a short question from the current conversation without analyzing source material.

How it works

The infranodus-mcp-server-infranodus MCP server exposes MCP tools that call InfraNodus graph and analysis workflows. Some tools return an analysis without necessarily creating a persistent graph, while graph-creation tools save the result and return its name and link for later use. Existing graphs can be discovered with list_graphs, searched with search, inspected with fetch, or analyzed by name.

For larger documents or existing corpora, generate_ontology_graph accepts one source: a prompt, text, or an existing graph name. It can produce either a codebase-oriented ontology or a procedural digest. Processing is chunked server-side, and the tool reports chunk progress. Ontologies and analytics can be returned together, while graph saving and analytics inclusion can be controlled through the tool options described by the server.

Analysis tools can identify content gaps, latent topics, conceptual bridges, research questions, and research ideas. optimize_text_structure evaluates bias and coherence in supplied text, while optimize_reasoning applies a related analysis to reasoning or conversation text. Retrieval tools use saved graph statements to provide context for later prompts.

Setup and configuration

The project is published on npm as infranodus-mcp-server. A local stdio installation can be started with:

npx -y infranodus-mcp-server

The README identifies Claude Desktop as a supported AI assistant, but the supplied material does not provide a complete client configuration block or document required environment variable names. Access to saved graphs is described in terms of the currently logged-in InfraNodus account, so account access may be needed for graph operations; the exact credential configuration is not specified here.

Tools and capabilities

The infranodus-mcp-server-infranodus MCP server includes tools for:

• Generating, creating, saving, and analyzing knowledge graphs.
• Building ontologies from prompts, long text, or existing graphs.
• Listing, searching, fetching, and retrieving information from saved graphs.
• Generating topical clusters, contextual hints, content gaps, research questions, and research ideas.
• Producing responses from graph context and storing or reading relations in InfraNodus memory.
• Comparing, merging, and finding differences between texts or graphs.
• Analyzing Google searches, YouTube results, related queries, and LLM-generated descriptions.
• Developing latent topics and conceptual bridges, or running the combined text-development workflow.

Limitations and notes

Several tools are intended specifically for source analysis. In particular, knowledge-graph generation should not be used for a short clarification that can already be answered from conversation context. Inputs and outputs vary by tool: some accept text, URLs, or graph names, while ontology generation requires exactly one of its supported source types.

The material does not state the service pricing, license, required API environment variables, or the authentication mechanism used by the npm process. It also does not establish compatibility with clients beyond the explicit Claude Desktop reference.