RedisProvider Class

Redis context provider with dynamic, filterable schema.

Stores context in Redis and retrieves scoped context. Uses full-text or optional hybrid vector search to ground model responses.

Create a Redis Context Provider.

Constructor

RedisProvider()

Parameters

Name	Description
redis_url	The Redis server URL. Default value: redis://localhost:6379
index_name	The name of the Redis index. Default value: context
prefix	The prefix for all keys in the Redis database. Default value: context
redis_vectorizer	The vectorizer to use for Redis. Default value: None
vector_field_name	The name of the vector field in Redis. Default value: None
vector_algorithm	The algorithm to use for vector search. Default value: None
vector_distance_metric	The distance metric to use for vector search. Default value: None
application_id	The application ID to scope the context. Default value: None
agent_id	The agent ID to scope the context. Default value: None
user_id	The user ID to scope the context. Default value: None
thread_id	The thread ID to scope the context. Default value: None
scope_to_per_operation_thread_id	Whether to scope to the per-operation thread ID. Default value: False
context_prompt	The context prompt to use for the provider. Default value: ## Memories Consider the following memories when answering user questions:
redis_index	The Redis index to use for the provider. Default value: None
overwrite_index	Whether to overwrite the existing Redis index. Default value: False

Methods

__init__	Create a Redis Context Provider.
__new__
invoked
invoking	Called before invoking the model to provide scoped context. Concatenates recent messages into a query, fetches matching memories from Redis. Prepends them as instructions.
search_all	Returns all documents in the index. Streams results via pagination to avoid excessive memory and response sizes.
thread_created	Called when a new thread is created. Captures the per-operation thread id when scoping is enabled to enforce single-thread usage.

__init__

Create a Redis Context Provider.

__init__(redis_url: str = 'redis://localhost:6379', index_name: str = 'context', prefix: str = 'context', redis_vectorizer: BaseVectorizer | None = None, vector_field_name: str | None = None, vector_algorithm: Literal['flat', 'hnsw'] | None = None, vector_distance_metric: Literal['cosine', 'ip', 'l2'] | None = None, application_id: str | None = None, agent_id: str | None = None, user_id: str | None = None, thread_id: str | None = None, scope_to_per_operation_thread_id: bool = False, context_prompt: str = ContextProvider.DEFAULT_CONTEXT_PROMPT, redis_index: Any = None, overwrite_index: bool = False)

Parameters

Name	Description
redis_url	str The Redis server URL. Default value: "redis://localhost:6379"
index_name	str The name of the Redis index. Default value: "context"
prefix	str The prefix for all keys in the Redis database. Default value: "context"
redis_vectorizer	<xref:redisvl.utils.vectorize.base.BaseVectorizer> | None The vectorizer to use for Redis. Default value: None
vector_field_name	str | None The name of the vector field in Redis. Default value: None
vector_algorithm	Literal['flat', 'hnsw'] | None The algorithm to use for vector search. Default value: None
vector_distance_metric	Literal['cosine', 'ip', 'l2'] | None The distance metric to use for vector search. Default value: None
application_id	str | None The application ID to scope the context. Default value: None
agent_id	str | None The agent ID to scope the context. Default value: None
user_id	str | None The user ID to scope the context. Default value: None
thread_id	str | None The thread ID to scope the context. Default value: None
scope_to_per_operation_thread_id	bool Whether to scope to the per-operation thread ID. Default value: False
context_prompt	str The context prompt to use for the provider. Default value: ContextProvider.DEFAULT_CONTEXT_PROMPT
redis_index	Any The Redis index to use for the provider. Default value: None
overwrite_index	bool Whether to overwrite the existing Redis index. Default value: False

__new__

__new__(**kwargs)

invoked

async invoked(request_messages: ChatMessage | Sequence[ChatMessage], response_messages: ChatMessage | Sequence[ChatMessage] | None = None, invoke_exception: Exception | None = None, **kwargs: Any) -> None

Parameters

Name	Description
request_messages Required	ChatMessage | Sequence[ChatMessage]
response_messages	ChatMessage | Sequence[ChatMessage] | None Default value: None
invoke_exception	Exception | None Default value: None
kwargs Required	Any

Returns

Type	Description
None

invoking

Called before invoking the model to provide scoped context.

Concatenates recent messages into a query, fetches matching memories from Redis. Prepends them as instructions.

async invoking(messages: ChatMessage | MutableSequence[ChatMessage], **kwargs: Any) -> Context

Parameters

Name	Description
messages Required	ChatMessage | MutableSequence[ChatMessage] List of new messages in the thread.
kwargs Required	Any

Keyword-Only Parameters

Name	Description
**kwargs	not used at present at present.

Returns

Type	Description
Context	Context object containing instructions with memories.

search_all

Returns all documents in the index.

Streams results via pagination to avoid excessive memory and response sizes.

async search_all(page_size: int = 200) -> list[dict[str, Any]]

Parameters

Name	Description
page_size	int Page size used for pagination under the hood. Default value: 200

Returns

Type	Description
list[dict[str, Any]]	List of all documents.

thread_created

Called when a new thread is created.

Captures the per-operation thread id when scoping is enabled to enforce single-thread usage.

async thread_created(thread_id: str | None) -> None

Parameters

Name	Description
thread_id Required	str | None The ID of the thread or None.

Returns

Type	Description
None

Attributes

schema_dict

Get the Redis schema dictionary, computing and caching it on first access.

DEFAULT_CONTEXT_PROMPT

DEFAULT_CONTEXT_PROMPT: Final[str] = '## Memories\nConsider the following memories when answering user questions:'