Skip to main content

MongoDB-RAG CLI Reference

MongoDB-RAG provides a command-line interface (CLI) to help manage vector search configurations, test indexes, and run searches efficiently.

📌 Usage​

To see available commands, run:

npx mongodb-rag --help

⚡ CLI Commands​

Initialize MongoRAG Configuration​

npx mongodb-rag init

Initializes the .mongodb-rag.json configuration file.

Test MongoDB Vector Search Index​

npx mongodb-rag test-index

Runs a test on the configured MongoDB Vector Search index.

Ingest a JSON File into MongoDB​

npx mongodb-rag ingest <file-path>
  • file-path: Path to a JSON file containing documents to ingest.

Create a MongoDB Atlas Vector Search Index​

npx mongodb-rag create-index

Creates a vector search index in MongoDB Atlas.

Delete a MongoDB Atlas Vector Search Index​

npx mongodb-rag delete-index

Deletes an existing vector search index.

Generate a Vector Embedding​

npx mongodb-rag generate-embedding <text>
  • text: The input text for which to generate an embedding.
npx mongodb-rag search <query>
  • query: The search query to retrieve semantically similar documents.

Ask a Question About Your Documents​

npx mongodb-rag ask <question>
  • question: Ask a natural language question and get an answer based on your documents.
  • Options:
    • --show-sources: Display the source documents used to generate the answer
    • --cite-sources: Include citations in the generated answer

Start an Interactive Chat Session​

npx mongodb-rag chat

Start an interactive chat session powered by your document knowledge base.

  • Options:
    • --show-sources: Display the source documents used to generate each response
    • --cite-sources: Include citations in the generated responses

Show All Indexes in the Configured MongoDB Collection​

npx mongodb-rag show-indexes

Displays all vector indexes available in the configured MongoDB collection.

Show Current MongoRAG Configuration​

npx mongodb-rag show-config

Displays the current .mongodb-rag.json configuration.

Edit MongoRAG Configuration​

npx mongodb-rag edit-config

Allows editing of the MongoRAG configuration file.

Clear MongoRAG Configuration​

npx mongodb-rag clear-config

Removes the current MongoRAG configuration.

Reset MongoRAG Configuration​

npx mongodb-rag reset-config

Resets the configuration by re-running the setup.

Create Environment File​

npx mongodb-rag create-env

Creates a .env file from your .mongodb-rag.json configuration. This command generates environment variables needed for your RAG application:

  • MONGODB_URI
  • EMBEDDING_PROVIDER
  • EMBEDDING_API_KEY
  • EMBEDDING_MODEL
  • VECTOR_INDEX
  • MONGODB_DATABASE
  • MONGODB_COLLECTION

🔍 Search vs Ask vs Chat - When to Use Each​

MongoDB-RAG offers three different ways to interact with your document knowledge base:

  • Search: Use when you need to see the raw documents that match your query

    npx mongodb-rag search "MongoDB vector search capabilities"

  • Ask: Use when you want a direct answer to a specific question

    npx mongodb-rag ask "What are the key features of MongoDB Atlas Vector Search?"

  • Chat: Use when you need an interactive conversation with context retention

    npx mongodb-rag chat --show-sources --cite-sources

🗣️ Conversational RAG Capabilities​

MongoDB-RAG goes beyond vector search to provide true conversational capabilities with your data:

  1. Natural Language Answers: Get complete, synthesized answers to your questions instead of just document snippets

  2. Context-Aware Chat: Have multi-turn conversations that reference your document knowledge base

  3. Provider Flexibility: Works with both OpenAI and Ollama for embedding and chat generation

  4. Source Attribution: Optionally cite document sources in responses with the --cite-sources flag

  5. Transparent Results: View the underlying documents with the --show-sources flag

For more details, refer to the MongoDB-RAG API Reference.