Changelog - [0.5.1] - 2025-10-27

New Features

Bug Fixes

1. Fix 0.5.1 version in pyproject.toml

Other Changes

Changelog - [0.5.0] - 2025-10-27

New Features

1. Support projecting specific relationship types for heterogeneous graphs.

Bug Fixes

1. Fix a bug where get_relationship_properties, get_node_properties, and get_node_labels might not returning entire answer set.
2. Fix a bug in the signature of nested arrays for article_rank, eigenvector_centrality, pagerank and k_nearest_neighbour that certain LLM clients do not support.

Other Changes

Changelog - [0.4.0] - 2025-09-18

New Features

Bug Fixes

1. Fix a bug where disconnected nodes were not projected.
2. Fix a bug where multiple nodes containing the identifying name could be nondeterministically matched. The user must now specify the exact name.
3. Support weighted degree centrality.
4. Support projecting specific node labels.
5. Fix a bug where projection would fail loading non-numbers relationship properties.
6. Fix duplicate logging.

Other Changes

Changelog - [0.3.0] - 2025-08-26

New Features

1. Add a new get_relationship_properties_keys tool.
2. Add targetNode filtering for longest_path.
3. Add support for similarity algorithms.
4. Support loading array node properties.

Bug Fixes

1. Return node names in several path algorithms that only returned node ids.
2. Fix a bug with loading node properties incorrectly.

Other Changes

Changelog - [0.2.0] - 2025-07-30

New Features

1. Project all (non-string) node properties using the appropriate (integer/float) types.
2. Support personalisation, post-filtering and node name mapping in centrality and community algorithms whenever appropriate.
3. Support post-filtering and node name mapping in community algorithms whenever appropriate.
4. Support orientation parameter in Degree Centrality.

Bug Fixes

1. Fix GDS calls for path and community algorithms and clean up returned data format.
2. Fix a bug during GDS graph projection that always created bidirectional relationships.

Other Changes

1. Removed minimum_weight_k_spanning_tree since it is write mode only, which may modify the database unexpectedly.

Changelog - [0.1.0] - 2025-07-14

New Features

• Initial release of the GDS Agent MCP Server
• Support for Neo4j Graph Data Science algorithms:
  • Centrality algorithms: Article Rank, Articulation Points, Betweenness Centrality, Bridges, CELF, Closeness Centrality, Degree Centrality, Eigenvector Centrality, PageRank, Harmonic Centrality, HITS
  • Path algorithms: Dijkstra's Shortest Path, Delta Stepping Shortest Path, Dijsktra's Single Source Shortest Path, A-Star, Yen's, Minimum Weight Spanning Tree, Minimum Weight k-Spanning Tree, Minimum Directed Steiner Tree, Prize Collecting Steiner Tree, All Pairs Shortest Path, Random Walk, BFS, DFS, Bellman-Ford Single Source Shortest Path, Longest Path
  • Community detection: Conductance, HDBSCAN, K-Core Decomposition, K-1 Coloring, K-Means Clustering, Label Propagation, Leiden, Local Clustering Coefficient, Louvain, Modularity Metric, Modularity Optimization, Strongly Connected Component, Triangle Count, Weakly Connected Component, Approximate Maximum K-Cut, Speaker-Listener Label Propagation
  • Similarity algorithms: Node similarity, Filtered Node Similarity, K-Nearest Neighbors, Filtered K-Nearest Neighbors
• Example London Underground dataset and loading script