{"id":"rileylemm-graphrag-mcp","name":"graphrag_mcp","homepage":null,"repo_url":"https://github.com/rileylemm/graphrag_mcp","category":"ai-ml","subcategories":[],"tags":["ai-ml","mcp","rag","graphdb","neo4j","vector-search","qdrant","retrieval","hybrid-search","python"],"what_it_does":"Provides an MCP (Model Context Protocol) server exposing tools/resources that query a hybrid GraphRAG datastore: Neo4j for graph-based context expansion and Qdrant for vector/semantic search, with hybrid retrieval combining both.","use_cases":["Hybrid semantic search over document chunks with graph relationship expansion","Question answering where relevant context should be expanded via Neo4j relationships","Retrieval-augmented generation pipelines for MCP-enabled clients (e.g., Claude Desktop/Cursor)","Exploring and retrieving from a GraphRAG schema and Qdrant collection metadata via MCP resources"],"not_for":["A hosted SaaS offering (it’s designed to run locally with your own DB instances)","Replacing dedicated Neo4j/Qdrant client SDKs for full administrative DB operations","Applications requiring strong user authentication/authorization for the MCP endpoint out of the box"],"best_when":"You control the runtime environment (local or internal network), and want LLMs to query an existing Neo4j+Qdrant GraphRAG index via MCP tools.","avoid_when":"You need multi-tenant security, public exposure without additional reverse-proxy protections, or you cannot ensure robust indexing/credential configuration.","alternatives":["Use Neo4j directly (with custom RAG logic) for graph expansion plus Qdrant for vector search","Build a REST/GraphQL API wrapper around Neo4j+Qdrant and call it from your LLM orchestration layer","Use Qdrant hybrid search plus external graph post-processing (application-side)","Use an existing GraphRAG framework integrated with your LLM tooling instead of MCP"],"af_score":55.5,"security_score":43.0,"reliability_score":32.5,"package_type":"mcp_server","discovery_source":["github"],"priority":"high","status":"evaluated","version_evaluated":null,"last_evaluated":"2026-03-30T15:19:52.570882+00:00","interface":{"has_rest_api":false,"has_graphql":false,"has_grpc":false,"has_mcp_server":true,"mcp_server_url":null,"has_sdk":false,"sdk_languages":[],"openapi_spec_url":null,"webhooks":false},"auth":{"methods":["Environment-variable based configuration for backend DB access (NEO4J_USER/NEO4J_PASSWORD)"],"oauth":false,"scopes":false,"notes":"No user-facing auth model is described for the MCP server itself; authentication is limited to connecting to Neo4j/Qdrant via your configured credentials/environment variables."},"pricing":{"model":null,"free_tier_exists":false,"free_tier_limits":null,"paid_tiers":[],"requires_credit_card":false,"estimated_workload_costs":null,"notes":"Self-hosted open-source (MIT). Costs are your infrastructure for Neo4j/Qdrant and compute (embeddings/retrieval)."},"requirements":{"requires_signup":false,"requires_credit_card":false,"domain_verification":false,"data_residency":[],"compliance":[],"min_contract":null},"agent_readiness":{"af_score":55.5,"security_score":43.0,"reliability_score":32.5,"mcp_server_quality":78.0,"documentation_accuracy":70.0,"error_message_quality":0.0,"error_message_notes":null,"auth_complexity":80.0,"rate_limit_clarity":10.0,"tls_enforcement":40.0,"auth_strength":35.0,"scope_granularity":20.0,"dependency_hygiene":55.0,"secret_handling":70.0,"security_notes":"TLS is not described (README only mentions bolt for Neo4j and localhost ports). Auth appears limited to database credentials via .env; there’s no described auth/authorization for the MCP server endpoint. Secrets are presumably kept in .env (not confirmed about logging), and no explicit guidance is provided about preventing prompt/tool injection, query sanitization, or enforcing least-privilege scopes between MCP client users and the backing databases.","uptime_documented":0.0,"version_stability":50.0,"breaking_changes_history":50.0,"error_recovery":30.0,"idempotency_support":"false","idempotency_notes":null,"pagination_style":"none","retry_guidance_documented":false,"known_agent_gotchas":["Server depends on local Neo4j and Qdrant being reachable at the configured host/ports; misconfiguration leads to tool failures.","Correct operation assumes the underlying GraphRAG index has been created in both Neo4j (graph data/relationships) and Qdrant (chunk embeddings).","No explicit guidance is provided about rate limits, pagination strategy for large result sets, or how tool outputs should be re-requested after partial failures."]}}