RagTool
Description
TheRagTool is designed to answer questions by leveraging the power of Retrieval-Augmented Generation (RAG) through CrewAI’s native RAG system.
It provides a dynamic knowledge base that can be queried to retrieve relevant information from various data sources.
This tool is particularly useful for applications that require access to a vast array of information and need to provide contextually relevant answers.
Example
The following example demonstrates how to initialize the tool and use it with different data sources:Code
Supported Data Sources
TheRagTool can be used with a wide variety of data sources, including:
- 📰 PDF files
- 📊 CSV files
- 📃 JSON files
- JSONL files (
.jsonland.ndjson) - 📝 Text
- 📁 Directories/Folders
- 🌐 HTML Web pages
- 📽️ YouTube Channels
- 📺 YouTube Videos
- 📚 Documentation websites
- 📝 MDX files
- 📄 DOCX files
- 🧾 XML files
- 📬 Gmail
- 📝 GitHub repositories
- 🐘 PostgreSQL databases
- 🐬 MySQL databases
- 🤖 Slack conversations
- 💬 Discord messages
- 🗨️ Discourse forums
- 📝 Substack newsletters
- 🐝 Beehiiv content
- 💾 Dropbox files
- 🖼️ Images
- ⚙️ Custom data sources
Parameters
TheRagTool accepts the following parameters:
- summarize: Optional. Whether to summarize the retrieved content. Default is
False. - adapter: Optional. A custom adapter for the knowledge base. If not provided, a CrewAIRagAdapter will be used.
- config: Optional. Configuration for the underlying CrewAI RAG system. Accepts a
RagToolConfigTypedDict with optionalembedding_model(ProviderSpec) andvectordb(VectorDbConfig) keys. All configuration values provided programmatically take precedence over environment variables.
Adding Content
You can add content to the knowledge base using theadd method:
Code
Agent Integration Example
Here’s how to integrate theRagTool with a CrewAI agent:
Code
Vector Database
RagTool stores content in a local Qdrant database by default, with OpenAI’s text-embedding-3-small as the embedder. To use ChromaDB, install it with pip install "crewai[chromadb]" and set "provider": "chromadb" in the vectordb config.
Content that an earlier version stored in ChromaDB is not read by the Qdrant default. Add it again, or set the
chromadb provider.Advanced Configuration
You can customize the behavior of theRagTool by providing a configuration dictionary:
Code
Embedding Model Configuration
Theembedding_model parameter accepts a crewai.rag.embeddings.types.ProviderSpec dictionary with the structure:
Supported Providers
These providers use ChromaDB and need
pip install "crewai[chromadb]": bedrock, cohere, google-generativeai, huggingface, instructor, jina, ollama, onnx, openclip, roboflow, sentence-transformer, text2vec. The others (such as openai, azure, openrouter, google-vertex, voyageai, watsonx, oracle and custom) work without it.OpenAI
OpenAI
main.py
api_key(str): OpenAI API keymodel_name(str): Model to use. Default:text-embedding-ada-002. Options:text-embedding-3-small,text-embedding-3-large,text-embedding-ada-002dimensions(int): Number of dimensions for the embeddingorganization_id(str): OpenAI organization IDapi_base(str): Custom API base URLapi_version(str): API versiondefault_headers(dict): Custom headers for API requests
OPENAI_API_KEYorEMBEDDINGS_OPENAI_API_KEY:api_keyOPENAI_ORGANIZATION_IDorEMBEDDINGS_OPENAI_ORGANIZATION_ID:organization_idOPENAI_MODEL_NAMEorEMBEDDINGS_OPENAI_MODEL_NAME:model_nameOPENAI_API_BASEorEMBEDDINGS_OPENAI_API_BASE:api_baseOPENAI_API_VERSIONorEMBEDDINGS_OPENAI_API_VERSION:api_versionOPENAI_DIMENSIONSorEMBEDDINGS_OPENAI_DIMENSIONS:dimensions
Cohere
Cohere
main.py
api_key(str): Cohere API keymodel_name(str): Model to use. Default:large. Options:embed-english-v3.0,embed-multilingual-v3.0,large,small
COHERE_API_KEYorEMBEDDINGS_COHERE_API_KEY:api_keyEMBEDDINGS_COHERE_MODEL_NAME:model_name
VoyageAI
VoyageAI
main.py
api_key(str): VoyageAI API keymodel(str): Model to use. Default:voyage-2. Options:voyage-3,voyage-3-lite,voyage-code-3,voyage-large-2input_type(str): Type of input. Options:document(for storage),query(for search)truncation(bool): Whether to truncate inputs that exceed max length. Default:Trueoutput_dtype(str): Output data typeoutput_dimension(int): Dimension of output embeddingsmax_retries(int): Maximum number of retry attempts. Default:0timeout(float): Request timeout in seconds
VOYAGEAI_API_KEYorEMBEDDINGS_VOYAGEAI_API_KEY:api_keyVOYAGEAI_MODELorEMBEDDINGS_VOYAGEAI_MODEL:modelVOYAGEAI_INPUT_TYPEorEMBEDDINGS_VOYAGEAI_INPUT_TYPE:input_typeVOYAGEAI_TRUNCATIONorEMBEDDINGS_VOYAGEAI_TRUNCATION:truncationVOYAGEAI_OUTPUT_DTYPEorEMBEDDINGS_VOYAGEAI_OUTPUT_DTYPE:output_dtypeVOYAGEAI_OUTPUT_DIMENSIONorEMBEDDINGS_VOYAGEAI_OUTPUT_DIMENSION:output_dimensionVOYAGEAI_MAX_RETRIESorEMBEDDINGS_VOYAGEAI_MAX_RETRIES:max_retriesVOYAGEAI_TIMEOUTorEMBEDDINGS_VOYAGEAI_TIMEOUT:timeout
Ollama
Ollama
main.py
model_name(str): Ollama model name (e.g.,llama2,mistral,nomic-embed-text)url(str): Ollama API endpoint URL. Default:http://localhost:11434/api/embeddings
OLLAMA_MODELorEMBEDDINGS_OLLAMA_MODEL:model_nameOLLAMA_URLorEMBEDDINGS_OLLAMA_URL:url
Amazon Bedrock
Amazon Bedrock
main.py
model_name(str): Bedrock model ID. Default:amazon.titan-embed-text-v1. Options:amazon.titan-embed-text-v1,amazon.titan-embed-text-v2:0,cohere.embed-english-v3,cohere.embed-multilingual-v3session(Any): Boto3 session object for AWS authentication
AWS_ACCESS_KEY_ID: AWS access keyAWS_SECRET_ACCESS_KEY: AWS secret keyAWS_REGION: AWS region (e.g.,us-east-1)
Azure OpenAI
Azure OpenAI
main.py
deployment_id(str): Required - Azure OpenAI deployment IDapi_key(str): Azure OpenAI API keyapi_base(str): Azure OpenAI resource endpointapi_version(str): API version. Example:2024-02-01model_name(str): Model name. Default:text-embedding-ada-002api_type(str): API type. Default:azuredimensions(int): Output dimensionsdefault_headers(dict): Custom headers
AZURE_OPENAI_API_KEYorEMBEDDINGS_AZURE_API_KEY:api_keyAZURE_OPENAI_ENDPOINTorEMBEDDINGS_AZURE_API_BASE:api_baseEMBEDDINGS_AZURE_DEPLOYMENT_ID:deployment_idEMBEDDINGS_AZURE_API_VERSION:api_versionEMBEDDINGS_AZURE_MODEL_NAME:model_nameEMBEDDINGS_AZURE_API_TYPE:api_typeEMBEDDINGS_AZURE_DIMENSIONS:dimensions
Google Generative AI
Google Generative AI
main.py
api_key(str): Google AI API keymodel_name(str): Model name. Default:gemini-embedding-001. Options:gemini-embedding-001,text-embedding-005,text-multilingual-embedding-002task_type(str): Task type for embeddings. Default:RETRIEVAL_DOCUMENT. Options:RETRIEVAL_DOCUMENT,RETRIEVAL_QUERY
GOOGLE_API_KEY,GEMINI_API_KEY, orEMBEDDINGS_GOOGLE_API_KEY:api_keyEMBEDDINGS_GOOGLE_GENERATIVE_AI_MODEL_NAME:model_nameEMBEDDINGS_GOOGLE_GENERATIVE_AI_TASK_TYPE:task_type
Google Vertex AI
Google Vertex AI
main.py
model_name(str): Model name. Default:textembedding-gecko. Options:text-embedding-004,textembedding-gecko,textembedding-gecko-multilingualproject_id(str): Google Cloud project ID. Default:cloud-large-language-modelsregion(str): Google Cloud region. Default:us-central1api_key(str): API key for authentication
GOOGLE_APPLICATION_CREDENTIALS: Path to service account JSON fileGOOGLE_CLOUD_PROJECTorEMBEDDINGS_GOOGLE_VERTEX_PROJECT_ID:project_idEMBEDDINGS_GOOGLE_VERTEX_MODEL_NAME:model_nameEMBEDDINGS_GOOGLE_VERTEX_REGION:regionEMBEDDINGS_GOOGLE_VERTEX_API_KEY:api_key
Jina AI
Jina AI
main.py
api_key(str): Jina AI API keymodel_name(str): Model name. Default:jina-embeddings-v2-base-en. Options:jina-embeddings-v3,jina-embeddings-v2-base-en,jina-embeddings-v2-small-en
JINA_API_KEYorEMBEDDINGS_JINA_API_KEY:api_keyEMBEDDINGS_JINA_MODEL_NAME:model_name
HuggingFace
HuggingFace
main.py
url(str): Full URL to HuggingFace inference API endpoint
HUGGINGFACE_URLorEMBEDDINGS_HUGGINGFACE_URL:url
Instructor
Instructor
main.py
model_name(str): HuggingFace model ID. Default:hkunlp/instructor-base. Options:hkunlp/instructor-xl,hkunlp/instructor-large,hkunlp/instructor-basedevice(str): Device to run on. Default:cpu. Options:cpu,cuda,mps,xpuinstruction(str): Instruction prefix for embeddings
EMBEDDINGS_INSTRUCTOR_MODEL_NAME:model_nameEMBEDDINGS_INSTRUCTOR_DEVICE:deviceEMBEDDINGS_INSTRUCTOR_INSTRUCTION:instruction
Sentence Transformer
Sentence Transformer
main.py
model_name(str): Sentence Transformers model name. Default:all-MiniLM-L6-v2. Options:all-mpnet-base-v2,all-MiniLM-L6-v2,paraphrase-multilingual-MiniLM-L12-v2device(str): Device to run on. Default:cpu. Options:cpu,cuda,mps,xpunormalize_embeddings(bool): Whether to normalize embeddings. Default:False
EMBEDDINGS_SENTENCE_TRANSFORMER_MODEL_NAME:model_nameEMBEDDINGS_SENTENCE_TRANSFORMER_DEVICE:deviceEMBEDDINGS_SENTENCE_TRANSFORMER_NORMALIZE_EMBEDDINGS:normalize_embeddings
ONNX
ONNX
main.py
preferred_providers(list[str]): List of ONNX execution providers in order of preference
EMBEDDINGS_ONNX_PREFERRED_PROVIDERS:preferred_providers(comma-separated list)
OpenCLIP
OpenCLIP
main.py
model_name(str): OpenCLIP model architecture. Default:ViT-B-32. Options:ViT-B-32,ViT-B-16,ViT-L-14checkpoint(str): Pretrained checkpoint name. Default:laion2b_s34b_b79k. Options:laion2b_s34b_b79k,laion400m_e32,openaidevice(str): Device to run on. Default:cpu. Options:cpu,cuda,mps,xpu
EMBEDDINGS_OPENCLIP_MODEL_NAME:model_nameEMBEDDINGS_OPENCLIP_CHECKPOINT:checkpointEMBEDDINGS_OPENCLIP_DEVICE:device
Text2Vec
Text2Vec
main.py
model_name(str): Text2Vec model name from HuggingFace. Default:shibing624/text2vec-base-chinese. Options:shibing624/text2vec-base-multilingual,shibing624/text2vec-base-chinese
EMBEDDINGS_TEXT2VEC_MODEL_NAME:model_name
Roboflow
Roboflow
main.py
api_key(str): Roboflow API key. Default:""(empty string)api_url(str): Roboflow inference API URL. Default:https://infer.roboflow.com
ROBOFLOW_API_KEYorEMBEDDINGS_ROBOFLOW_API_KEY:api_keyROBOFLOW_API_URLorEMBEDDINGS_ROBOFLOW_API_URL:api_url
WatsonX (IBM)
WatsonX (IBM)
main.py
model_id(str): WatsonX model identifierurl(str): WatsonX API endpointapi_key(str): IBM Cloud API keyproject_id(str): WatsonX project IDspace_id(str): WatsonX space ID (alternative to project_id)batch_size(int): Batch size for embeddings. Default:100concurrency_limit(int): Maximum concurrent requests. Default:10persistent_connection(bool): Use persistent connections. Default:True- Plus 20+ additional authentication and configuration options
WATSONX_API_KEYorEMBEDDINGS_WATSONX_API_KEY:api_keyWATSONX_URLorEMBEDDINGS_WATSONX_URL:urlWATSONX_PROJECT_IDorEMBEDDINGS_WATSONX_PROJECT_ID:project_idEMBEDDINGS_WATSONX_MODEL_ID:model_idEMBEDDINGS_WATSONX_SPACE_ID:space_idEMBEDDINGS_WATSONX_BATCH_SIZE:batch_sizeEMBEDDINGS_WATSONX_CONCURRENCY_LIMIT:concurrency_limitEMBEDDINGS_WATSONX_PERSISTENT_CONNECTION:persistent_connection
Custom
Custom
main.py
embedding_callable(type[EmbeddingFunction]): Custom embedding function class
EmbeddingFunction protocol defined in crewai.rag.core.base_embeddings_callable. The __call__ method should accept input data and return embeddings as a list of numpy arrays (or compatible format that will be normalized). The returned embeddings are automatically normalized and validated.Notes
- All config fields are optional unless marked as Required
- API keys can typically be provided via environment variables instead of config
- Default values are shown where applicable
Conclusion
TheRagTool provides a powerful way to create and query knowledge bases from various data sources. By leveraging Retrieval-Augmented Generation, it enables agents to access and retrieve relevant information efficiently, enhancing their ability to provide accurate and contextually appropriate responses.