Skip to main content
Stores provide a secure and consistent way to manage credentials and sensitive data across different storage backends. Use stores to set, retrieve, and delete credentials in a unified manner, regardless of the underlying storage mechanism.

​
Features

  • Abstract base class defining a consistent interface for credential stores
  • Multiple storage backend implementations:
    • In-memory store (temporary storage)
    • Environment variable store
    • System keyring store (secure credential storage)
  • Exception handling for common error cases
  • Type hints and comprehensive documentation

​
When to Use Each Store

Choose the store type that best fits your environment and security needs:
  • MemoryStore: Use for testing or temporary storage. Data is lost when the program exits.
  • EnvironmentStore: Ideal for CI/CD pipelines or containerized environments where environment variables are managed externally.
  • KeyringStore: Recommended for production and local development when you need secure, persistent storage using the system keyring.

​
Available Store Types

​
MemoryStore

A simple in-memory store that persists data only for the duration of program execution. Useful for testing or temporary storage. 
from universal_mcp.stores import MemoryStore

store = MemoryStore()
store.set("api_key", "secret123")
value = store.get("api_key")  # Returns "secret123"

​
EnvironmentStore

Uses environment variables to store and retrieve credentials. Useful for containerized environments or CI/CD pipelines. 
from universal_mcp.stores import EnvironmentStore

store = EnvironmentStore()
store.set("API_KEY", "secret123")
value = store.get("API_KEY")  # Returns "secret123"

​
KeyringStore

Leverages the system’s secure credential storage facility. Provides the most secure option for storing sensitive data. 
from universal_mcp.stores import KeyringStore

store = KeyringStore(app_name="my_app")
store.set("api_key", "secret123")
value = store.get("api_key")  # Returns "secret123"

​
Getting Started

Here’s how to use a store to set and retrieve a credential: 
from universal_mcp.stores import MemoryStore

store = MemoryStore()
store.set("api_key", "secret123")
value = store.get("api_key")  # Returns "secret123"
Switch to EnvironmentStore or KeyringStore by importing and instantiating the relevant class.

​
Best Practices

  1. Use KeyringStore for production environments where security is a priority.
  2. Use EnvironmentStore for containerized or cloud environments.
  3. Use MemoryStore for testing or temporary storage only.
  4. Always handle StoreError and KeyNotFoundError exceptions appropriately.

​
Error Handling

The store module provides specific exceptions:
  • StoreError: Base exception for all store-related errors.
  • KeyNotFoundError: Raised when a requested key is not found.
Handle these exceptions to ensure robust credential management: 
from universal_mcp.stores import KeyNotFoundError

try:
    value = store.get("missing_key")
except KeyNotFoundError:
    print("Credential not found.")

​
Dependencies

  • keyring: Required for the KeyringStore implementation
  • loguru: Used for logging operations in the KeyringStore

​
Contributing

To add a new store implementation, inherit from BaseStore and implement all required abstract methods:
  • get(key: str) -> Any
  • set(key: str, value: str) -> None
  • delete(key: str) -> None
For a comprehensive guide, see the Universal MCP Stores Reference below.