Skip to main content

Available Tools

The Knowledge Tree MCP server exposes 8 read-only tools for navigating the knowledge graph.

search_graph

Search the knowledge graph for nodes matching a text query.

Parameters:

NameTypeRequiredDefaultDescription
querystringyesSearch term for concept names
limitintno20Max results (1-100)
node_typestringnoFilter: concept, entity, perspective, event

Returns: List of matching nodes with node_id, concept, node_type, fact_count, and also_known_as aliases.

Example:

search_graph(query="climate change", limit=5, node_type="concept")

get_node

Load a node's core details including definition, parent, creation date, and counts.

Parameters:

NameTypeRequiredDescription
node_idstringyesUUID of the node

Returns: Node details including concept, node_type, definition, parent_id, parent_concept, fact_count, edge_count, dimension_count, aliases, merged_from, and metadata. If the node has no definition, a fallback dimension is included.

Example:

get_node(node_id="a1b2c3d4-...")

get_dimensions

Load dimensions (model perspectives) for a node. Each dimension is a different AI model's independent analysis.

Parameters:

NameTypeRequiredDefaultDescription
node_idstringyesUUID of the node
limitintno10Max dimensions (1-50)
offsetintno0Skip N dimensions

Returns: Array of dimensions with model_id, content, confidence, and generated_at. Includes pagination info (returned, total, offset).


get_edges

Load edges (relationships) for a node, sorted by fact count (most evidence first).

Parameters:

NameTypeRequiredDefaultDescription
node_idstringyesUUID of the node
limitintno10Max edges (1-100)
offsetintno0Skip N edges
edge_typestringnoFilter: related or cross_type

Returns: Array of edges with other_node_id, other_concept, other_node_type, relationship_type, weight, justification, and fact_count. Sorted by evidence strength.


get_facts

Load facts linked to a node, grouped by source. Supports powerful filtering for targeted research.

Parameters:

NameTypeRequiredDefaultDescription
node_idstringyesUUID of the subject node
limitintno50Max facts (1-200)
offsetintno0Skip N facts
source_node_idstringnoUUID of a second node — returns only facts linked to BOTH nodes
author_orgstringnoFilter by author organization (partial match)
source_domainstringnoFilter by source URL domain (partial match)
searchstringnoFilter by fact content text
fact_typestringnoFilter: claim, account, measurement, quote, etc.

Returns: Facts organized by source groups, each containing uri, title, author_org, published_date, and an array of facts. Includes next_offset for pagination.

Node intersection filtering

The source_node_id parameter is the most powerful filter. It returns only facts linked to both the primary node and the source node. This is the best way to answer questions like "What does CNN say about topic X?" — pass the topic as node_id and "CNN" as source_node_id.

Example:

get_facts(
node_id="<epstein-node-uuid>",
source_node_id="<cnn-node-uuid>"
)

get_fact_sources

Load deduplicated sources for a node's facts — useful for understanding provenance.

Parameters:

NameTypeRequiredDescription
node_idstringyesUUID of the node

Returns: List of unique sources with uri, title, provider_id, published_date, author_person, author_org, and fact_count.


search_facts

Search the global fact pool by text query, optionally scoped to a node.

Parameters:

NameTypeRequiredDefaultDescription
querystringnoText search query
node_idstringnoScope search to a specific node's facts
limitintno20Max results (1-100)
offsetintno0Skip N results
fact_typestringnoFilter by fact type
author_orgstringnoFilter by author organization
source_domainstringnoFilter by source domain

Returns: Array of facts with content, fact_type, source information, and provenance details.


get_node_paths

Find shortest paths between two nodes in the graph using breadth-first search.

Parameters:

NameTypeRequiredDefaultDescription
source_node_idstringyesUUID of the start node
target_node_idstringyesUUID of the end node
max_depthintno4Maximum path length
limitintno3Maximum number of paths to return

Returns: Array of paths, each containing an ordered list of nodes from source to target with their concepts and types.