Skip to content

Config reference: graph_storage

  • Enterprise tuning surface

    Defaults + constraints are rendered directly from Pydantic.

  • Env keys when available

    Many fields have an env-style alias (from TriBridConfig.to_flat_dict()).

  • Tooltip-level guidance

    If a matching glossary entry exists, you’ll see deeper tuning notes.

Config reference Config API & workflow Glossary

Total parameters: 14

Group index
  • (root)

(root)

JSON key Env key(s) Type Default Constraints Summary
graph_storage.community_algorithm GRAPH_COMMUNITY_ALGORITHM Literal["louvain", "label_propagation"] "louvain" allowed="louvain", "label_propagation" Community detection algorithm
graph_storage.entity_types — list[str] ["function", "class", "module", "variable", "import"] — Entity types to extract and store in graph
graph_storage.graph_search_top_k GRAPH_SEARCH_TOP_K int 30 ≥ 5, ≤ 100 Number of results from graph traversal
graph_storage.include_communities GRAPH_INCLUDE_COMMUNITIES bool true — Include community detection in graph analysis
graph_storage.max_hops GRAPH_MAX_HOPS int 2 ≥ 1, ≤ 5 Maximum traversal hops for graph search
graph_storage.neo4j_auto_create_databases — bool true — Automatically create per-corpus Neo4j databases when missing (Enterprise).
graph_storage.neo4j_database NEO4J_DATABASE str "neo4j" — Neo4j database name
graph_storage.neo4j_database_mode — Literal["shared", "per_corpus"] "shared" allowed="shared", "per_corpus" Database isolation mode: 'shared' uses a single Neo4j database (Community-compatible), 'per_corpus' uses a separate Neo4j database per corpus (Enterprise multi-database).
graph_storage.neo4j_database_prefix — str "tribrid_" — Prefix for per-corpus Neo4j database names when neo4j_database_mode='per_corpus'.
graph_storage.neo4j_password NEO4J_PASSWORD str "" — Neo4j password (defaults to NEO4J_PASSWORD env var when unset)
graph_storage.neo4j_uri NEO4J_URI str "bolt://localhost:7687" — Neo4j connection URI (bolt:// or neo4j://)
graph_storage.neo4j_user NEO4J_USER str "neo4j" — Neo4j username
graph_storage.neo4j_vector_query_mode NEO4J_VECTOR_QUERY_MODE Literal["auto", "procedure", "search"] "auto" allowed="auto", "procedure", "search" Neo4j chunk-vector query mode. 'auto' prefers runtime-safe defaults and only uses SEARCH where supported.
graph_storage.relationship_types — list[str] ["calls", "imports", "inherits", "contains", "references"] — Relationship types to extract

Details (glossary)

graph_storage.graph_search_top_k (GRAPH_SEARCH_TOP_K) — Graph Search Top-K

Category: general

GRAPH_SEARCH_TOP_K controls how many graph candidates are kept before downstream fusion and reranking. Increasing top-k usually improves recall because more potentially useful graph evidence survives early pruning, but it also raises latency and can inflate reranker and generation token costs. If top-k is too small, graph retrieval appears weak even when the graph is high quality because relevant nodes are dropped prematurely. If it is too large, weaker graph neighbors can crowd the context budget and reduce final answer precision. Tune this setting with both retrieval metrics and end-to-end answer quality, and keep it aligned with final context assembly limits.

Badges: - Top-K Control

Links: - Neo4j GraphRAG Python User Guide - Neo4j Vector Indexes - Elasticsearch Similarity and Ranking - LightRetriever (2025): Faster Query Inference

graph_storage.include_communities (GRAPH_INCLUDE_COMMUNITIES) — Include Communities

Category: general

GRAPH_INCLUDE_COMMUNITIES enables expansion across precomputed graph communities instead of only direct neighbors. This can surface related components that belong to the same subsystem even when explicit edges between the exact seed nodes are weak or missing. It is most useful for architecture, ownership, and impact-analysis questions where thematic grouping matters. The tradeoff is broader recall with higher risk of topic drift, so community expansion should usually be combined with conservative hop limits and robust reranking. Community quality depends heavily on graph construction and algorithm settings, so treat this as a quality-dependent feature flag rather than always-on behavior.

Badges: - Advanced Graph

Links: - Neo4j Louvain Algorithm - Neo4j Leiden Algorithm - Neo4j GraphRAG Python User Guide - TagRAG (2026): Tag-Guided Hierarchical GraphRAG

graph_storage.max_hops (GRAPH_MAX_HOPS) — Graph Max Hops

Category: general

GRAPH_MAX_HOPS caps traversal depth from each seed node in graph retrieval. One hop focuses on direct relationships, two hops often captures practical cross-file links, and larger values rapidly increase branching factor and latency. Higher hops can help for dependency-chain and architecture questions, but they also raise the chance of pulling weakly related evidence into the fusion stage. In most RAG/search deployments, this is one of the highest-impact latency controls because frontier size grows nonlinearly with graph degree. Tune with p95 latency and grounded answer metrics together, since deeper traversal can improve recall while reducing precision.

Badges: - Latency-Recall

Links: - Cypher Variable-Length Patterns - Neo4j GraphRAG Python User Guide - Neo4j Cypher Query Tuning - TagRAG (2026): Tag-Guided Hierarchical GraphRAG

graph_storage.neo4j_uri (NEO4J_URI) — Neo4j Connection URI

Category: infrastructure

Neo4j URI config determines how clients connect, route, and secure graph queries in retrieval workflows. Use neo4j:// for routed cluster-aware connections and bolt:// for direct connections when routing is not needed. Misconfigured schemes can produce subtle behavior differences in failover, read routing, and TLS handling that only appear under load. Treat this value as infrastructure configuration: validate connectivity at startup, enforce encrypted transport in shared environments, and keep URI/auth settings externalized from source code.

Links: - SCOUT-RAG: Dynamic Graph Retrieval-Augmented Generation (arXiv 2026) - Neo4j Browser DBMS Connection - Neo4j Python Driver Advanced Connection - Neo4j GraphRAG