Getting Started

Configuration

Complete environment variables reference for DocuStore.

DocuStore is configured via environment variables. All settings are defined in services/.env.

LLM Provider

Variable	Description	Default
`LLM_PROVIDER`	Provider type: `ollama`, `openai`, or `gemini`	`ollama`
`LLM_MODEL`	Model name (e.g., `llama3.1`, `gpt-4o`)	—
`LLM_BASE_URL`	Provider API base URL	—
`OPENAI_API_KEY`	OpenAI API key (if using OpenAI)	—
`GEMINI_API_KEY`	Gemini API key (if using Gemini)	—

Embedding

Variable	Description	Default
`EMBEDDING_MODEL`	Embedding model name	`nomic-embed-text-v1.5`
`EMBEDDING_DIMENSIONS`	Vector dimensions	`768`
`EMBEDDING_DEVICE`	Compute device: `cpu` or `cuda`	`cpu`

Storage

Variable	Description	Default
`BLOB_STORE_TYPE`	`local` or `s3`	`local`
`BLOB_STORE_PATH`	Local storage path	`./blobs`
`S3_BUCKET`	S3 bucket name (if using S3)	—
`S3_REGION`	S3 region	—

Infrastructure

Variable	Description	Default
`ESDB_URI`	EventStoreDB connection string	`esdb://eventstore:2113?tls=false`
`MONGODB_URI`	MongoDB connection string	`mongodb://mongo:27017`
`MONGODB_DATABASE`	MongoDB database name	`docu_store`
`QDRANT_HOST`	Qdrant hostname	`qdrant`
`QDRANT_PORT`	Qdrant gRPC port	`6334`
`TEMPORAL_HOST`	Temporal server address	`temporal:7233`
`KAFKA_BOOTSTRAP_SERVERS`	Kafka broker addresses	`kafka:9092`

Temporal Workers

Variable	Description	Default
`TEMPORAL_MAX_CONCURRENT_ACTIVITIES`	Max parallel activities per worker	`3`
`TEMPORAL_TASK_QUEUE`	Task queue name	`docu-store-tasks`

Observability

Variable	Description	Default
`LANGFUSE_PUBLIC_KEY`	Langfuse public key for LLM tracing	—
`LANGFUSE_SECRET_KEY`	Langfuse secret key	—
`LANGFUSE_HOST`	Langfuse server URL	—

Summarization

Variable	Description	Default
`ARTIFACT_SUMMARIZATION_BATCH_SIZE`	Pages per summarization batch	`10`

Vector Collections

Variable	Description	Default
`QDRANT_COLLECTION_NAME`	Text embeddings collection	`text_embeddings`
`QDRANT_COMPOUND_COLLECTION_NAME`	SMILES embeddings collection	`compound_embeddings`
`QDRANT_SUMMARY_COLLECTION_NAME`	Summary embeddings collection	`summary_embeddings`

Quickstart

Deploy DocuStore with Docker Compose and process your first document.

Production Deployment

Deploy DocuStore to production with Docker Swarm.