n8n-nodes-lightrag

n8n community node for the LightRAG knowledge graph API. Full coverage of the LightRAG server API v1.4.11+ with 20 operations across 5 resources.

Install

Community Node (recommended)

1. In n8n, go to Settings > Community Nodes
2. Enter n8n-nodes-lightrag
3. Click Install

Manual

cd ~/.n8n/nodes
npm install n8n-nodes-lightrag
# Restart n8n

Credentials

| Field | Description | |-------|-------------| | Base URL | Your LightRAG server URL (e.g. http://localhost:9621) | | API Key | X-API-Key header value. Leave empty if auth is disabled. |

The credential test calls GET /health to verify connectivity and authentication.

Operations

Document

| Operation | Description | Endpoint | |-----------|-------------|----------| | Insert Text | Index a single text document | POST /documents/text | | Insert Texts | Index multiple texts in one call | POST /documents/texts | | Upload File | Upload a file for indexing | POST /documents/upload | | List | List documents with status filter and pagination | POST /documents/paginated | | Get Status | Poll processing status by track ID | GET /documents/track_status/{id} | | Delete | Delete documents by ID | DELETE /documents/delete_document |

Query

| Operation | Description | Endpoint | |-----------|-------------|----------| | Query | RAG query with LLM-generated response | POST /query | | Query Data | Raw data retrieval (entities, relations, chunks) without LLM | POST /query/data |

Query modes: mix (adaptive), local (entity-focused), global (graph traversal), hybrid (local + global), naive (text search), bypass (direct LLM).

Additional options: top_k, chunk_top_k, include_references, response_type, conversation_history, hl_keywords, ll_keywords, max_total_tokens, only_need_context, only_need_prompt, user_prompt.

Graph

| Operation | Description | Endpoint | |-----------|-------------|----------| | List Labels | List all entity labels in the graph | GET /graph/label/list | | Popular Entities | Get most-connected entities | GET /graph/label/popular | | Search Labels | Fuzzy search entity labels | GET /graph/label/search | | Get Subgraph | Get neighborhood graph around an entity | GET /graphs | | Check Entity | Check if an entity exists | GET /graph/entity/exists | | Create Entity | Create a new entity node | POST /graph/entity/create | | Edit Entity | Modify entity properties | POST /graph/entity/edit | | Merge Entities | Merge duplicate entities into one | POST /graph/entities/merge |

Relation

| Operation | Description | Endpoint | |-----------|-------------|----------| | Create | Create a relationship between two entities | POST /graph/relation/create | | Edit | Update relationship properties | POST /graph/relation/edit |

System

| Operation | Description | Endpoint | |-----------|-------------|----------| | Health Check | Get server health, config, and storage info | GET /health |

Batch Processing

All operations are batch-aware. When multiple items flow into the node, each item is processed independently with its own parameters. Use n8n expressions (e.g. {{ $json.query }}) to pass dynamic values per item.

Example Workflows

Insert + Poll Status

[Read File] → [LightRAG: Upload File] → [Wait 5s] → [LightRAG: Get Status]

Query + Process Results

[Trigger] → [LightRAG: Query] → [IF has references] → [Process references]

Graph Exploration

[LightRAG: Search Labels] → [LightRAG: Get Subgraph] → [Visualize]

Compatibility

• n8n: >= 1.0.0
• LightRAG server: >= 1.4.11
• Node.js: 18, 20, 22

Development

git clone https://github.com/menezis-ai/n8n-nodes-lightrag.git
cd n8n-nodes-lightrag
npm install
npm run build       # Compile TypeScript
npm run lint        # Type check
npm run dev         # Watch mode

License

MIT