gatecraft 0.1.1

A semantic RBAC system with RAG capabilities

Gatecraft

Gatecraft is a semantic Role-Based Access Control (RBAC) system with Retrieval-Augmented Generation (RAG) capabilities. It provides a flexible way to manage access control based on semantic similarity and content understanding.

Features

• Semantic-based access control
• Integration with OpenAI embeddings
• Pinecone vector store support
• Flexible RBAC system
• RAG capabilities

Installation

pip install gatecraft

Quick Start

Here's a quick example to get you started with Gatecraft:

from gatecraft import Gatecraft, SemanticCondition

# Initialize the Gatecraft system
gc = Gatecraft()

# Create users
alice = gc.create_user(user_id=1, name='Alice')
bob = gc.create_user(user_id=2, name='Bob')

# Create roles with conditions
cat_lover_role = gc.create_role(role_id=1, name='Cat Lover')
cat_condition = SemanticCondition(term='cat', threshold=0.78)
gc.add_condition_to_role(cat_lover_role, cat_condition)

everything_except_cats_role = gc.create_role(role_id=2, name='Everything Except Cats')
inverse_cat_condition = SemanticCondition(term='cat', threshold=0.78, inverse=True)
gc.add_condition_to_role(everything_except_cats_role, inverse_cat_condition)

# Assign roles to users
gc.assign_role(alice, cat_lover_role)
gc.assign_role(bob, everything_except_cats_role)

# Add entities (documents)
gc.add_entity(entity_id=1, data='Cute kitten playing with a ball of yarn.')
gc.add_entity(entity_id=2, data='Dog training and obedience tips.')

# Check access
print(f"Alice access to Entity 1: {gc.is_access_allowed(alice, 1)}")  # Should be True
print(f"Alice access to Entity 2: {gc.is_access_allowed(alice, 2)}")  # Should be False
print(f"Bob access to Entity 1: {gc.is_access_allowed(bob, 1)}")      # Should be False
print(f"Bob access to Entity 2: {gc.is_access_allowed(bob, 2)}")      # Should be True

Explanation

• Initialization: Start by importing Gatecraft and SemanticCondition from the gatecraft library. Initialize the Gatecraft system:

  from gatecraft import Gatecraft, SemanticCondition
  
  gc = Gatecraft()
  

• Creating Users: Create user instances for Alice and Bob:

  alice = gc.create_user(user_id=1, name='Alice')
  bob = gc.create_user(user_id=2, name='Bob')
  

• Defining Roles and Conditions:

  • Cat Lover Role: Create a role for users who love cats and add a condition to allow access to content related to "cat":

    cat_lover_role = gc.create_role(role_id=1, name='Cat Lover')
    cat_condition = SemanticCondition(term='cat', threshold=0.78)
    gc.add_condition_to_role(cat_lover_role, cat_condition)
    

  • Everything Except Cats Role: Create a role for users who should access all content except that related to "cat":

    everything_except_cats_role = gc.create_role(role_id=2, name='Everything Except Cats')
    inverse_cat_condition = SemanticCondition(term='cat', threshold=0.78, inverse=True)
    gc.add_condition_to_role(everything_except_cats_role, inverse_cat_condition)
    

• Assigning Roles to Users: Assign the roles to the respective users:

  gc.assign_role(alice, cat_lover_role)
  gc.assign_role(bob, everything_except_cats_role)
  

• Adding Entities (Documents): Add content entities to the system:

  gc.add_entity(entity_id=1, data='Cute kitten playing with a ball of yarn.')
  gc.add_entity(entity_id=2, data='Dog training and obedience tips.')
  

• Access Control Check: Determine if users have access to specific entities:

  print(f"Alice access to Entity 1: {gc.is_access_allowed(alice, 1)}")  # True
  print(f"Alice access to Entity 2: {gc.is_access_allowed(alice, 2)}")  # False
  print(f"Bob access to Entity 1: {gc.is_access_allowed(bob, 1)}")      # False
  print(f"Bob access to Entity 2: {gc.is_access_allowed(bob, 2)}")      # True
  

Additional Information

• Semantic Conditions: The SemanticCondition allows you to define access conditions based on semantic similarity to a term. Setting inverse=True reverses the condition.
• Similarity Threshold: Adjust the threshold parameter in SemanticCondition to change the sensitivity of the semantic matching.
• Gatecraft Class: The Gatecraft class encapsulates all the functionality, simplifying the process of creating users, roles, and entities, and checking access.
• No Need to Manage Permissions Directly: Permissions are handled internally, reducing complexity and making the library easier to use.

Advanced Usage

For more advanced scenarios, you can:

• Integrate with Different Vector Stores: Provide a custom vector store to the Gatecraft class if you wish to use something other than Pinecone.

  from gatecraft.db.mock_vector_store import MockVectorStore
  
  vector_store = MockVectorStore()
  gc = Gatecraft(vector_store=vector_store)
  

• Set Custom Similarity Thresholds: Adjust the overall similarity threshold when initializing Gatecraft:

  gc = Gatecraft(similarity_threshold=0.90)
  

Dependencies

Ensure that you have the required API keys set in your environment variables or .env file:

• OPENAI_API_KEY
• PINECONE_API_KEY
• PINECONE_ENVIRONMENT

Contributing

Contributions are welcome! Please open an issue or submit a pull request on GitHub.

License

This project is licensed under the MIT License - see the LICENSE file for details.

---

Feel free to reach out if you have any questions or need further assistance with Gatecraft!