Epstein File Explorer MCP Server

Search and explore publicly released documents related to Jeffrey Epstein from official government sources, court records, FOIA releases, and verified archives.

Tools

• get_stats

  Get overview statistics for the Epstein document collection and knowledge graph. Returns counts of files, pages, entities, relationships, collections, extraction progress, search availability, and last-updated timestamp. Call this first to understand the scope of the dataset.

• search_entities

  Search entities by name, type, or sub-label. Args: query: Search term to match against entity names and aliases (case-insensitive substring match) entity_type: Filter by entity type (Person, Organization, Location, Property, Aircraft, LegalCase, etc.) sub_label: Filter by sub-label (Victim, Defendant, Financier, GovernmentAgency, etc.) limit: Maximum results to return (default 25, max 100) offset: Number of results to skip for pagination (default 0)

• get_entity

  Get full details for a specific entity including connections, documents, and evidence. Args: name: The canonical name of the entity (e.g., "Jeffrey Epstein", "Virginia Giuffre") Returns entity details with all aliases, sub-labels, top connections (with relationship types), and documents where the entity appears.

• search_relationships

  Search relationships between entities. Args: relationship_type: Filter by relationship type (e.g., FLEW_ON, EMPLOYED_BY, PAID, DEFENDANT_IN) domain: Filter by relationship domain (financial, legal, social, abuse, power, event, employment, property, communication, coverup, document) entity_name: Filter to relationships involving this entity (source or target) limit: Maximum results (default 25, max 100) offset: Number of results to skip for pagination

• get_document_entities

  Get all entities mentioned in a specific document. Args: file_id: The document file ID (get IDs from search_documents) Returns the document info and a list of entities mentioned in it with their types and mention counts.

• search_documents

  Search documents in the collection. Args: query: Search term to match against filenames (case-insensitive substring) collection: Filter by collection name (e.g., "bondi-phase-1", "data-set-06", "fbi-vault") file_type: Filter by file type (pdf, mp4, wav, etc.) limit: Maximum results (default 25, max 100) offset: Number of results to skip for pagination (default 0) Returns dict with 'items' list, 'total' count, 'offset', 'limit', and 'has_more' fla

• get_entity_network

  Get the relationship network around an entity up to a specified depth. Args: name: The canonical name of the center entity depth: How many hops to traverse (1 = direct connections, 2 = connections of connections). Max 2. Returns a network with nodes (entities) and edges (relationships) suitable for understanding the entity's position in the knowledge graph.

• read_document_text

  Read the extracted text content of a document. Args: file_id: The document file ID max_chars: Maximum characters to return (default 5000, max 50000) char_start: Start reading from this character offset (default 0). Useful for deep-linking to specific evidence. char_end: Stop reading at this character offset (default 0 = use max_chars from char_start). Returns the document text content (if text extraction was performed).

• search_text

  Full-text search across all extracted document content. Searches the text content of all documents for a keyword or phrase. Uses FTS5 full-text index when available, falling back to disk scan. Args: query: Text to search for (case-insensitive). Can be a name, phrase, or keyword. collection: Optionally restrict to a specific collection (e.g., "bondi-phase-1", "data-set-06") limit: Maximum documents to return (default 10, max 50) Returns documents contain

• get_entity_evidence

  Get all evidence (mentions and relationships) for an entity, with verification scores and deep-links. Args: name: The canonical name of the entity (e.g., "Jeffrey Epstein", "Virginia Giuffre") file_id: Optionally filter to evidence from a specific document (file ID) limit: Maximum results per category (default 30, max 100) offset: Number of results to skip per category for pagination (default 0) Returns entity mentions and relationships with deep-links t

• get_relationship_evidence

  Get all evidence supporting a relationship between two entities. Args: source: Source entity name target: Target entity name relationship_type: Optionally filter by relationship type (e.g., FLEW_ON, ABUSED_BY) limit: Maximum results (default 20, max 50) offset: Number of results to skip for pagination (default 0) Returns dict with 'items' list, 'total' count, 'offset', 'limit', and 'has_more' flag. Use offset += limit to fetch the next page w

• get_verification_stats

  Get quality and verification statistics for the knowledge graph. Returns verification status distributions, position match rates, confidence distributions, and per-collection/domain quality metrics. Useful for understanding data quality.

• get_collections

  Get summary statistics for each document collection. Returns per-collection counts of files, total bytes, total pages, and file types. Useful for understanding the scope and composition of each data source.

• get_entity_facets

  Get entity type distribution and sub-label facets for filtering. Returns counts of entities by type (Person, Organization, etc.) and counts by sub-label (Victim, Defendant, etc.). Useful for understanding entity composition and building search filters.

• get_entity_connections

  Get simplified list of connected entities for a given entity. Args: name: The canonical name of the entity limit: Maximum connections to return (default 50, max 200) Returns a flat list of connected entities with relationship type, domain, confidence, and optional deep-links. Simpler than get_entity for quick lookups.

• search_relationship_text

  Full-text search across relationship evidence text. Searches relationship descriptions, evidence quotes, and context for a keyword or phrase. Uses FTS5 index (rel_search table) when available. Args: query: Text to search for (min 2 chars). Can be a name, phrase, or keyword. domain: Filter by relationship domain (financial, legal, social, abuse, etc.) limit: Maximum results (default 20, max 50) Returns matching relationships with source/target entities,

• aggregate_query

  Run aggregate analytics queries over the knowledge graph. Args: metric: The analytics query to run. One of: - top_entities: entities ranked by relationship count - relationship_distribution: count of relationships by type - entity_type_distribution: count of entities by type - collection_coverage: entities appearing across the most document collections - top_connected: entities with highest degree centrality (most distinct

• find_path

  Find connection paths between two entities in the knowledge graph. Uses bidirectional breadth-first search to efficiently find how two entities are connected through the relationship network. Args: source: Source entity name (e.g., "Leslie Wexner") target: Target entity name (e.g., "Virginia Giuffre") max_depth: Maximum hops to search (default 4, max 6) Returns up to 5 shortest paths between the entities, each with relationship hops showing entity n

• compare_entities

  Compare two entities side-by-side: shared connections, shared documents, and network overlap. Useful for investigative analysis — find common associates, co-occurrences in documents, and the degree of overlap between two entities' networks. Args: entity_a: First entity name (e.g., "Ghislaine Maxwell") entity_b: Second entity name (e.g., "Jean-Luc Brunel") Returns comparison data including shared connections (with relationship types for each entity), shared