InfraNodus MCP Server
InfraNodus MCP server turns any text into a knowledge graph and shows topic clusters, the gaps between them and research questions
- 104 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- Texts and links go to the InfraNodus API and graphs are saved in your account
- The delete_graph, delete_statements and update_statements tools change graphs irreversibly, but run only after confirmation
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add mcp/infranodus-mcp-serverDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- Texts and links go to the InfraNodus API and graphs are saved in your account
- The delete_graph, delete_statements and update_statements tools change graphs irreversibly, but run only after confirmation
This is third-party code. Review the repository files before installing.
What it does
The server connects an AI agent to an InfraNodus account, a service for text network analysis. The agent passes a text, a link or a YouTube transcript, and the server builds a concept graph, finds topical clusters and shows which topics are weakly connected or not connected at all. From those gaps the tools generate research questions, ideas and bridges between topics. Separate tools compare two texts and find their overlap or difference, check a text for bias and coherence, analyse the model's own reasoning and build an ontology for a topic or a codebase. Graphs can be saved in the account, searched and used as agent memory: the server writes entities and relations, then retrieves them on request. It runs locally through npx or remotely at https://mcp.infranodus.com with OAuth sign-in.
Who it is for. Researchers, SEO specialists and writers who work through long texts with Claude and want to see the structure of a topic and what it leaves out.
Good fit when
- You have a long interview transcript or a big article and want the main topics in a minute without manual tagging
- An SEO content plan is built on topical clusters and needs keyword groups with a short summary of each
- A research draft feels one-sided and you want questions that sit between weakly connected topics
- Two documents need comparing to find what the first lacks and the second has
- The model's reasoning in a long session goes in circles and you want it checked for bias and coherence
Not a fit when
- A short summary of the text is enough: a graph and clusters are overkill for that
- The document is confidential and cannot go to an external service: the server calls the InfraNodus API
- You need exact phrase search inside documents: a concept graph does not replace full-text search
- You have no InfraNodus account and the 70 free requests without a key are already used
Example request
Build a graph from this interview transcript, show the main topical clusters and suggest three questions on the gaps between loosely connected clustersMCP
- Transport
- stdio, http
- Authentication
- key is optional
Environment variables
INFRANODUS_API_KEYsecret- API key from the InfraNodus account; without it 70 requests work
INFRANODUS_DELETE- A value of 0 removes the graph delete and edit tools
INFRANODUS_FEEDBACK- A value of 0 turns off workflow feedback reports
Limitations
- Without an API key the server answers 70 requests, after that an InfraNodus account is needed; the service offers a 14-day trial.
- Three tools change graphs in your account: delete_statements, update_statements and delete_graph. The changes are irreversible, but a dry run comes first and they execute only on a repeat call with confirm. Setting INFRANODUS_DELETE=0 removes these tools.
- The submit_workflow_feedback tool sends the developers short reports on how a workflow went (a task paraphrase of up to 200 characters). INFRANODUS_FEEDBACK=0 turns it off.
- Running locally requires Node.js 18 or newer.
How to disable. Run claude mcp remove infranodus to take the server out of Claude Code. Graphs in the InfraNodus account stay: delete them in the service itself.
FAQ
How do I connect InfraNodus to Claude Code?
Run claude mcp add infranodus -s user -- env INFRANODUS_API_KEY=your_key npx -y infranodus-mcp-server. The key comes from the api-access page in your InfraNodus account. For web clients the remote address https://mcp.infranodus.com is simpler: the client prompts for OAuth sign-in on its own.
Do I need an InfraNodus account for the server to work?
Not for first tests: without a key the server handles 70 requests. After that an account and an API key are needed, and the service has a 14-day trial. The agent sees your saved graphs only with a key.
Can the agent delete my graphs?
It can, but not at once. The delete_graph, delete_statements and update_statements tools first show what would change and apply it after a repeat call with confirmation. If you do not want deletion rights, set INFRANODUS_DELETE=0 and those three tools disappear from the server.
How do I use InfraNodus as agent memory?
Ask the agent to save memories into a specific graph, for example my-memories. The server writes entities and relations as a graph and retrieves them on request. You can limit extraction, for instance to people and companies only.
What does the generate_content_gaps tool do?
It looks for missing links between topic clusters in a text and returns underdeveloped topics, research questions and ideas for developing the material. The result is built on the graph structure and not on a retelling of the text.
Discussion
No questions or reviews yet.
Related
MCP serversA knowledge graph with an MCP server: the agent searches concepts by meaning and sees how well each is supported.
MCP server for the Flexible GraphRAG platform: ingest from 14 data sources, hybrid search and AI answers over a knowledge graph
The official Graphlit MCP server: an agent collects material from Slack, GitHub, mail and websites into a knowledge base and searches it
A CLI and MCP server that extracts structured knowledge from documents: graphs, hypergraphs, and fact search with sources
An MCP server with a toolbox for global briefs: markets, SEC filings, conflict, climate and news from open sources
A local MCP server that turns a folder of documents into a knowledge base with hybrid search for Claude Code and Cursor
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Authentication | Repository stars | Main difference |
|---|---|---|---|---|
| InfraNodus MCP Server (this card) | yes | key is optional | 104 | |
| MCP Knowledge Graph | no | not required | 888 | keeps entities and relations in local files as agent memory, with no analysis of gaps between topics |
| Cognee | no | 33,093 | builds agent memory on a knowledge graph, while InfraNodus analyses the structure of the text itself |
README in short
The README opens with an overview and a list of tools with a description of each, then covers setup: the remote address with OAuth, npx launch, building from source and Smithery. It continues with Claude Desktop configurations for macOS and Windows, example requests, advice on using the graph as memory and a section on project notes for the agent.