MCP SERVER · OFFICIAL PROJECTBuilds knowledge graphs from text, URLs or YouTube videos, then finds the gaps in the discourse and generates questions to fill them.
LABSCO SUMMARY
Every tool here can tell you what a document says. InfraNodus tells you what it conspicuously does not — which topics sit unconnected, and what question would bridge them. For research, editorial planning or competitive analysis that is the more valuable output, and it comes from graph structure rather than from asking a model to speculate. Auto-transcribing a YouTube URL into the same pipeline is a practical touch.
InfraNodus' endpoint for their text-network analysis product. The distinctive tools are not about building graphs but about reading what is missing from them.
OAuth 2.0 bearer, header only — and the consent step is unusual: you paste your InfraNodus API key on their auth page, so OAuth wraps an API-key handoff. Local deployments use an environment variable instead, with anonymous use allowed for 70 requests.
Paste a URL, then authorize — add the endpoint to your client, then approve the OAuth consent screen
USE CASES
Build the graph, then read the content gaps and the questions that would bridge them.
Pass the YouTube URL; it transcribes and graphs the content.
Generate responses from the graph's shape rather than from a flat retrieval.
Save it in InfraNodus and analyse it again later by name.
Listed in project docs. Your client may show a different list.
generate_knowledge_graphGenerate a knowledge graph from text, URL (incl. YouTube video, auto-transcribed), or an existing InfraNodus graph, with full structural analysis: nodes, relations, topical clusters, gaps, diversity statistics.
create_knowledge_graphCreate a knowledge graph from text or URL and SAVE it in InfraNodus for future reference / as a knowledge base. Returns same as generate_knowledge_graph plus graphName and graphUrl.
generate_ontology_graphAI-generate a reasoning ontology (entities + relations) for a topic/prompt/text and by default save it as an InfraNodus graph. numberOfResults = approx number of ontology relations (max 40, default 40); saveGraph:false returns ontologyStatements instead of graph links.
generate_topical_clustersExtract topical clusters from text, URL, YouTube video or existing graph. Compact delivery of the most important topics.
generate_content_gapsIdentify areas of discourse that could be further developed by linking topics that are not well connected.
generate_research_questionsGenerate research questions that bridge the content gaps found in a text.
generate_research_ideasGenerate ideas to develop text further; bridges content gaps within the text's context, or with shouldTranscend:true focuses on least-represented clusters and conceptual gateways to link to a wider discourse.
generate_responses_from_graphGenerate prompt-based answers and expert advice synthesized from a graph's structure (as opposed to verbatim retrieval, which retrieve_from_knowledge_base does).
generate_contextual_hintStructural summary for GraphRAG augmentation — returns a textOverview string with XML-like tags for MainConcepts, MainTopics, TopicalGaps, ConceptualGateways, Relations, DiversityStatistics.
analyze_existing_graph_by_nameExtract an existing InfraNodus graph and provide full graph analysis: clusters, concepts, gaps, diversity statistics.
analyze_textGeneral text analysis from text, URL or YouTube video, focused on analysis results and structural recommendations (diversity/focus). Includes an analysed statements array.
develop_latent_topicsIdentify underdeveloped topics and generate ideas (requestMode:'transcend') or research questions to develop them.
develop_conceptual_bridgesLike develop_latent_topics but focused on nodes with a high betweenness-to-degree ratio — concepts that link different topical clusters ('outside the box' thinking).
develop_text_toolComprehensive text development combining optimize / latent / conceptual-bridges modes; transcendDiscourse:true pushes beyond the text.
optimize_text_structureAnalyze bias and coherence: if too biased develop least-represented topics; if focused/diversified develop content gaps; if dispersed develop most common gap topics. responseType:'transcend' connects to wider context.
optimize_reasoningListed as tool #10 in the official repository README (github.com/infranodus/mcp-server-infranodus). NOT present in the website tools catalog at https://infranodus.com/mcp/tools.
list_graphsList all graphs in the user's account; searchable by name, type, date, language, favorites. Returns totalGraphs + graph records (id, name, type, isFavorite, createdAt, language). Present in the website catalog; NOT in the repository README list.
retrieve_from_knowledge_baseGraphRAG + RAG retrieval: get an existing graph by name and retrieve statements relevant to a prompt (semantic — e.g. 'sin' retrieves relevant Bible-graph content). includeGraphSummary:true adds a contextual overview.
searchFind all statements in the user's account containing a search term. Vendor notes this tool is REQUIRED for MCP connections to ChatGPT.
fetchFetch the statements found by the search tool using the ID it returned. Required alongside search for ChatGPT connections.
memory_add_relationsAdd relations to a memory graph (creates the graph if absent). Entities are detected as [[wikilinks]] nodes by default, producing a high-level concept representation.
memory_get_relationsRetrieve memory statements containing a specific entity, or all statements in a memory graph when entity is empty.
difference_between_textsShow what is missing in the first text/URL/graph relative to the others — missing relations and clusters, not just keywords.
overlap_between_textsFind common topics and relations present in ALL provided texts/URLs/graphs.
merged_graph_from_textsGenerate a merged graph from several texts, URLs and existing graphs — overview of a discourse across sources.
analyze_google_search_resultsBuild a graph of Google search results for given queries — which topics to cover for topical authority. includeSearchResults:true returns the retrieved URLs.
analyze_youtube_resultsListed as tool #22 in the official repository README, with controls limit (default 100, max 2000), sortBy (Popular/Oldest/Latest), excludeDescriptions, importLanguage, importRegion. NOT present in the website tools catalog.
analyze_related_search_queriesAnalyze which search queries people use together with the provided ones (Google 'people also search for'), optionally with search volume.
search_queries_vs_search_resultsFind high-volume search-query clusters that do NOT appear in Google search results — demand/supply content gaps. includeSearchQueries:true returns actual queries with volume.
analyze_llm_resultsAsk an LLM to describe a topic, then graph its response to reveal how the model frames it (model bias / LLMO / GEO). numberOfResults = number of separate LLM completions (default 20, each a model call). includeAnalytics defaults true, includeGraph defaults false.
generate_seo_reportFull SEO report: extract text+keywords, retrieve search results (topical authority) and search intent (demand), derive what people search for but don't find, synthesize content ideas and gap recommendations. Execution may take 60-90 seconds; emits MCP progress notifications at 6 checkpoints.
get_more_toolsMETA / progressive disclosure. 'Checks for additional specialized InfraNodus tools whenever a task might benefit from capabilities not already loaded. Some MCP clients expose a reduced toolset by default; this lets the LLM request the rest on demand.' Returns additional tool definitions that then become callable. Present in the website catalog only.
enable_project_learningsCreate the opt-in, per-project, append-only learnings graph (learn-<project>) in your account, where the assistant saves what it learned about operating in a project. Called only when you explicitly ask; idempotent.
add_project_learningsSave learnings about a project (where things live, traps, conventions, decisions, workflows) as typed statements for later sessions. Refuses when the project has not been enabled; dry run by default, writes only with confirm: true.
get_project_learningsRetrieve learnings for a project by prompt (GraphRAG), by entity (a file path, module or concept), or as a structural overview; with no project, lists the projects that have learnings.
optimize_knowledge_baseStructural feedback on a whole code base, document vault or body of procedural knowledge from its digest graph: what dominates, which areas are under-developed, which clusters never connect, with suggestions for what to develop next.
submit_workflow_feedbackInternal telemetry: after a workflow of InfraNodus calls, the assistant reports what it did with the output; it never asks you anything and never surfaces in the conversation.
delete_statementsDelete statements from a graph in your own account by one filter (categories, exact statements, query, date window, deleteAll or statementIds). Dry run by default; nothing is removed until the same filter is sent with confirm: true. Irreversible.
update_statementsEdit statements of a graph in your own account in place (content, categories or timestamp), by exact text or id, or in bulk by selector with set / replace. Dry run by default; writes with confirm: true. Irreversible.
delete_graphDelete one graph from your own account entirely: its statements, revisions, settings and URL. Dry run by default; nothing is removed until the same graphName is sent with confirm: true. Irreversible.
BEFORE YOU INSTALL
40 tools named in the project's documentation.
an MCP client that can reach a remote endpoint
INSTALLATION
Add the endpoint to your client, then approve the OAuth consent screen.
https://mcp.infranodus.comMCP servers
MattArtzAnthro
baobab-tech
Mistizz
Osseni94