Skip to content

cayu.memory.processing

Public Cayu API. Definitions are grouped by their public source module, and definitions from private implementation modules appear on their package’s page; adding a symbol does not move existing links. The import shown at each entry is a supported public namespace. Read the Python API overview and install the required extra.

from cayu.memory import AGENT_RECALL_PROCESSING_SCHEMA_VERSION
AGENT_RECALL_PROCESSING_SCHEMA_VERSION = 'cayu.agent_recall_processing.v3'

Release source · Runtime guarantees · Capability guides

from cayu.memory import agent_recall_situation_input_sha256

Fingerprint retrieval-shaping input, excluding clock and access authority.

agent_recall_situation_input_sha256(situation: RecallSituation) -> str

Release source · Runtime guarantees · Capability guides

from cayu.memory import agent_work_context_recall_text

Render deterministic bounded retrieval text from durable work context.

agent_work_context_recall_text(context: AgentWorkContext) -> str

Release source · Runtime guarantees · Capability guides

from cayu.memory import AgentRecallFrontier

A caller-replayable pair of unchanged-context delta high-water marks.

Bases: _AgentRecallModel.

Defaults below are source expressions; Field(...) includes schema/validation settings and default_factory creates a fresh value. Inherited declarations are included. Constructor behavior follows the model or dataclass shown in the bases/decorators.

Name Type Default
knowledge_sequence int required
index_readiness_sequence int required
AgentRecallFrontier.model_copy(*, update: Mapping[str, Any] | None=None, deep: bool=False) -> Self
AgentRecallFrontier.validate_sequence(value: int, info) -> int
AgentRecallFrontier.fingerprint() -> str

Release source · Runtime guarantees · Capability guides

from cayu.memory import AgentRecallProcessingError

A required freshness source could not be evaluated safely.

Bases: RuntimeError.

AgentRecallProcessingError.__init__(code: str) -> None

Release source · Runtime guarantees · Capability guides

from cayu.memory import AgentRecallProcessingMode

Bases: StrEnum.

Defaults below are source expressions; Field(...) includes schema/validation settings and default_factory creates a fresh value. Inherited declarations are included. Constructor behavior follows the model or dataclass shown in the bases/decorators.

Name Type Default
FULL_INDEX class attribute 'full_index'
DELTA class attribute 'delta'
NO_WORK class attribute 'no_work'

Release source · Runtime guarantees · Capability guides

from cayu.memory import AgentRecallProcessingRequest

One side-effect-free full-index or delta recall request.

Bases: _AgentRecallModel.

Defaults below are source expressions; Field(...) includes schema/validation settings and default_factory creates a fresh value. Inherited declarations are included. Constructor behavior follows the model or dataclass shown in the bases/decorators.

Name Type Default
agent_id str required
work_context AgentWorkContext required
situation RecallSituation required
checkpoint_stream_id str required
checkpoint AgentRecallCheckpoint | None None
frontier AgentRecallFrontier | None None
processing_id str required
operation_id str required
updated_by str required
updated_at datetime required
AgentRecallProcessingRequest.model_copy(*, update: Mapping[str, Any] | None=None, deep: bool=False) -> Self
AgentRecallProcessingRequest.validate_identity(value: str, info) -> str
AgentRecallProcessingRequest.copy_work_context(value: object, info) -> AgentWorkContext
AgentRecallProcessingRequest.copy_situation(value: object, info) -> RecallSituation
AgentRecallProcessingRequest.copy_checkpoint(value: object, info) -> AgentRecallCheckpoint | None
AgentRecallProcessingRequest.copy_frontier(value: object) -> AgentRecallFrontier | None
AgentRecallProcessingRequest.validate_updated_at(value: datetime) -> datetime
AgentRecallProcessingRequest.validate_authority() -> AgentRecallProcessingRequest

Release source · Runtime guarantees · Capability guides

from cayu.memory import AgentRecallProcessingResult

Immutable retrieval outcome and optional, unapplied checkpoint proposal.

Bases: _AgentRecallModel.

Defaults below are source expressions; Field(...) includes schema/validation settings and default_factory creates a fresh value. Inherited declarations are included. Constructor behavior follows the model or dataclass shown in the bases/decorators.

Name Type Default
schema_version Literal['cayu.agent_recall_processing.v3'] AGENT_RECALL_PROCESSING_SCHEMA_VERSION
mode AgentRecallProcessingMode required
reason str required
agent_id str required
task_id str required
knowledge_namespace str required
work_context_revision int required
work_context_sha256 str required
processing_id str required
operation_id str required
access_policy_sha256 str required
checkpoint_stream_id str required
situation_sha256 str required
frontier AgentRecallFrontier required
processed_frontier AgentRecallFrontier required
knowledge_event_count int required
index_readiness_event_count int required
knowledge_events tuple[KnowledgeChange, ...] ()
index_readiness_events tuple[KnowledgeIndexReadiness, ...] ()
eligible_revisions tuple[KnowledgeRevisionRef, ...] ()
recall RecallResult | None None
proposed_checkpoint AgentRecallCheckpoint | None None
frontier_complete bool required
retry_required bool False
AgentRecallProcessingResult.model_copy(*, update: Mapping[str, Any] | None=None, deep: bool=False) -> Self
AgentRecallProcessingResult.validate_mode(value: object) -> AgentRecallProcessingMode
AgentRecallProcessingResult.validate_identity(value: str, info) -> str
AgentRecallProcessingResult.validate_sha256(value: str, info) -> str
AgentRecallProcessingResult.validate_work_context_revision(value: int) -> int
AgentRecallProcessingResult.validate_work_context_sha256(value: str) -> str
AgentRecallProcessingResult.validate_event_count(value: int, info) -> int
AgentRecallProcessingResult.copy_knowledge_events(value) -> tuple[KnowledgeChange, ...]
AgentRecallProcessingResult.copy_index_readiness_events(value) -> tuple[KnowledgeIndexReadiness, ...]
AgentRecallProcessingResult.copy_revisions(value) -> tuple[KnowledgeRevisionRef, ...]
AgentRecallProcessingResult.validate_boolean(value: object, info) -> bool
AgentRecallProcessingResult.validate_result_shape() -> AgentRecallProcessingResult
AgentRecallProcessingResult.fingerprint() -> str

Release source · Runtime guarantees · Capability guides

from cayu.memory import AgentRecallProcessor

Select and evaluate a full-index, delta, or explicit no-work recall lane.

AgentRecallProcessor.__init__(knowledge_store: KnowledgeStore, *, fusion_config: WeightedReciprocalRankFusionConfig, engine_config: RecallEngineConfig | None=None, config: AgentRecallProcessorConfig | None=None) -> None
async AgentRecallProcessor.process(request: AgentRecallProcessingRequest) -> AgentRecallProcessingResult

Release source · Runtime guarantees · Capability guides

from cayu.memory import AgentRecallProcessorConfig

Hard work and output bounds for one processor invocation.

Bases: _AgentRecallModel.

Defaults below are source expressions; Field(...) includes schema/validation settings and default_factory creates a fresh value. Inherited declarations are included. Constructor behavior follows the model or dataclass shown in the bases/decorators.

Name Type Default
knowledge_change_limit int 100
index_readiness_limit int 100
candidate_limit int 20
source_max_bytes int 64000
max_record_bytes int 8000
semantic_timeout_seconds float 1.0
lineage_limit int 0
lineage_candidate_limit int 10
lineage_max_bytes int 8192
AgentRecallProcessorConfig.model_copy(*, update: Mapping[str, Any] | None=None, deep: bool=False) -> Self
AgentRecallProcessorConfig.validate_page_limit(value: int, info) -> int
AgentRecallProcessorConfig.validate_candidate_limit(value: int) -> int
AgentRecallProcessorConfig.validate_result_byte_bound(value: int, info) -> int
AgentRecallProcessorConfig.validate_lineage_limit(value: int) -> int
AgentRecallProcessorConfig.validate_lineage_candidate_limit(value: int) -> int
AgentRecallProcessorConfig.validate_lineage_max_bytes(value: int) -> int
AgentRecallProcessorConfig.validate_semantic_timeout(value: object) -> float
AgentRecallProcessorConfig.validate_combined_revision_bound() -> AgentRecallProcessorConfig

Release source · Runtime guarantees · Capability guides