@falkordb/mcpserver
Model Context Protocol server for FalkorDB graph databases - enables AI assistants to query and manage graph data using natural language
Versions
1.3.0latest1.2.31.2.21.2.11.2.0+ show 3 moreshow less
1.1.11.1.01.0.1Tools 7
query_graph_readonly Run a read-only OpenCypher query on a graph using GRAPH.RO_QUERY. This ensures no write operations are performed and is ideal for replica instances.
list_graphs List all graphs available to query
delete_graph Permanently delete a graph from the database. WARNING: This action is irreversible. You must set confirmDelete to true to proceed.
query_graph Run an OpenCypher query on a graph. Supports both read-write and read-only queries.
get_graph_schema Get the schema of a graph including node labels, relationship types, and (optionally) the connection topology between them, to understand the graph structure before executing a query. Connection topology is derived from a bounded sample of relationships and can be disabled via includeConnections on very large graphs.
get_node_schema Sample up to N nodes of a given label and aggregate their property keys by how many nodes carry each one (descending). Interpret each property's frequency relative to the returned sampledCount (the actual number of nodes sampled, which may be less than requestedSampleSize) to detect naming drift — e.g. a property present on only a few of many sampled nodes likely duplicates another under a slightly different name. Especially valuable in schemaless graphs where property naming can drift across subsets of nodes.
get_relationship_schema Sample up to N relationships of a given type and aggregate their property keys by how many relationships carry each one (descending). Interpret each property's frequency relative to the returned sampledCount (the actual number of relationships sampled, which may be less than requestedSampleSize) to detect naming drift — e.g. a property present on only a few of many sampled relationships likely duplicates another under a slightly different name. Especially valuable in schemaless graphs where property naming can drift across subsets of relationships.
Permissions 1
env_vars low