cayu.evals.store

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.

`EVAL_RUN_INVOCATION_MAX_BYTES`

from cayu.evals import EVAL_RUN_INVOCATION_MAX_BYTES

EVAL_RUN_INVOCATION_MAX_BYTES = 64 << 10

Release source · Runtime guarantees · Capability guides

`EVAL_RUN_MAX_OBSERVATION_INTERVAL_SECONDS`

from cayu.evals import EVAL_RUN_MAX_OBSERVATION_INTERVAL_SECONDS

EVAL_RUN_MAX_OBSERVATION_INTERVAL_SECONDS = 1.0

Release source · Runtime guarantees · Capability guides

`EVAL_RUN_MAX_TERMINAL_WAIT_SECONDS`

from cayu.evals import EVAL_RUN_MAX_TERMINAL_WAIT_SECONDS

EVAL_RUN_MAX_TERMINAL_WAIT_SECONDS = 300.0

Release source · Runtime guarantees · Capability guides

`EVAL_RUN_MIN_OBSERVATION_INTERVAL_SECONDS`

from cayu.evals import EVAL_RUN_MIN_OBSERVATION_INTERVAL_SECONDS

EVAL_RUN_MIN_OBSERVATION_INTERVAL_SECONDS = 0.05

Release source · Runtime guarantees · Capability guides

`eval_run_observation`

from cayu.evals import eval_run_observation

Project a full public run record into its lightweight lifecycle shape.

eval_run_observation(record: EvalRunRecord) -> EvalRunObservation

Release source · Runtime guarantees · Capability guides

`EVAL_STORE_DEFAULT_PAGE_BYTES`

from cayu.evals import EVAL_STORE_DEFAULT_PAGE_BYTES

EVAL_STORE_DEFAULT_PAGE_BYTES = 1 << 20

Release source · Runtime guarantees · Capability guides

`EVAL_STORE_DEFAULT_PAGE_SIZE`

from cayu.evals import EVAL_STORE_DEFAULT_PAGE_SIZE

EVAL_STORE_DEFAULT_PAGE_SIZE = 50

Release source · Runtime guarantees · Capability guides

`EVAL_STORE_MAX_CURSOR_BYTES`

from cayu.evals import EVAL_STORE_MAX_CURSOR_BYTES

EVAL_STORE_MAX_CURSOR_BYTES = 1024

Release source · Runtime guarantees · Capability guides

`EVAL_STORE_MAX_LEASE_SECONDS`

from cayu.evals import EVAL_STORE_MAX_LEASE_SECONDS

EVAL_STORE_MAX_LEASE_SECONDS = 3600

Release source · Runtime guarantees · Capability guides

`EVAL_STORE_MAX_PAGE_BYTES`

from cayu.evals import EVAL_STORE_MAX_PAGE_BYTES

EVAL_STORE_MAX_PAGE_BYTES = 8 << 20

Release source · Runtime guarantees · Capability guides

`EVAL_STORE_MAX_PAGE_SIZE`

from cayu.evals import EVAL_STORE_MAX_PAGE_SIZE

EVAL_STORE_MAX_PAGE_SIZE = 200

Release source · Runtime guarantees · Capability guides

`EvalAuthoredSuiteCatalogEntry`

from cayu.evals import EvalAuthoredSuiteCatalogEntry

Bases: _EvalStoreModel.

Fields and attributes

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
`revision`	`StrictStr`	`required`
`id`	`StrictStr`	`required`
`suite_revision`	`StrictStr`	`required`
`target_key`	`StrictStr`	`required`
`name`	`StrictStr`	`required`
`description`	`StrictStr \| None`	`None`
`case_count`	`StrictInt`	`Field(ge=1, le=EVAL_CORPUS_MAX_CASES)`
`assertion_count`	`StrictInt`	`Field(ge=1, le=EVAL_CORPUS_MAX_CASES * EVAL_CORPUS_MAX_ASSERTIONS_PER_CASE)`
`simple_input_count`	`StrictInt`	`Field(ge=0, le=EVAL_CORPUS_MAX_CASES)`
`scenario_count`	`StrictInt`	`Field(ge=0, le=EVAL_CORPUS_MAX_CASES)`
`trials`	`StrictInt`	`Field(ge=1, le=EVAL_CORPUS_MAX_TRIALS)`
`timeout_seconds`	`StrictInt`	`Field(ge=1, le=EVAL_CORPUS_MAX_TIMEOUT_SECONDS)`
`document_bytes`	`StrictInt`	`Field(ge=1, le=EVAL_SUITE_AUTHORING_MAX_BYTES)`
`created_at`	`datetime`	`required`

validate_revisions (method)

EvalAuthoredSuiteCatalogEntry.validate_revisions(value: str, info) -> str

validate_ids (method)

EvalAuthoredSuiteCatalogEntry.validate_ids(value: str, info) -> str

validate_name (method)

EvalAuthoredSuiteCatalogEntry.validate_name(value: str, info) -> str

validate_description (method)

EvalAuthoredSuiteCatalogEntry.validate_description(value: str | None, info) -> str | None

validate_created_at (method)

EvalAuthoredSuiteCatalogEntry.validate_created_at(value: datetime, info) -> datetime

validate_counts (method)

EvalAuthoredSuiteCatalogEntry.validate_counts() -> EvalAuthoredSuiteCatalogEntry

Release source · Runtime guarantees · Capability guides

`EvalAuthoredSuiteCatalogPage`

from cayu.evals import EvalAuthoredSuiteCatalogPage

Bases: _EvalStoreModel.

Fields and attributes

Name	Type	Default
`items`	`tuple[EvalAuthoredSuiteCatalogEntry, ...]`	`Field(max_length=EVAL_STORE_MAX_PAGE_SIZE)`
`next_cursor`	`StrictStr \| None`	`None`
`has_more`	`StrictBool`	`False`

validate_page (method)

EvalAuthoredSuiteCatalogPage.validate_page() -> EvalAuthoredSuiteCatalogPage

Release source · Runtime guarantees · Capability guides

`EvalAuthoredSuiteCatalogQuery`

from cayu.evals import EvalAuthoredSuiteCatalogQuery

Bases: _EvalStoreModel.

Fields and attributes

Name	Type	Default
`target_key`	`StrictStr \| None`	`None`
`suite_id`	`StrictStr \| None`	`None`
`cursor`	`StrictStr \| None`	`None`
`limit`	`StrictInt`	`Field(default=EVAL_STORE_DEFAULT_PAGE_SIZE, ge=1, le=EVAL_STORE_MAX_PAGE_SIZE)`
`max_result_bytes`	`StrictInt`	`Field(default=EVAL_STORE_DEFAULT_PAGE_BYTES, ge=1024, le=EVAL_STORE_MAX_PAGE_BYTES)`

validate_ids (method)

EvalAuthoredSuiteCatalogQuery.validate_ids(value: str | None, info) -> str | None

validate_cursor (method)

EvalAuthoredSuiteCatalogQuery.validate_cursor(value: str | None, info) -> str | None

Release source · Runtime guarantees · Capability guides

`EvalAuthoredSuiteConflict`

from cayu.evals import EvalAuthoredSuiteConflict

One immutable authored-suite revision resolves to contradictory content.

Bases: ValueError.

This type’s implementation and inherited behavior are available in the linked release source.

Release source · Runtime guarantees · Capability guides

`EvalAuthoredSuiteReferenceError`

from cayu.evals import EvalAuthoredSuiteReferenceError

An authored suite references an unavailable or incompatible scenario.

Bases: ValueError.

This type’s implementation and inherited behavior are available in the linked release source.

Release source · Runtime guarantees · Capability guides

`EvalBaselineConflict`

from cayu.evals import EvalBaselineConflict

A baseline mutation lost its compare-and-swap or idempotency contract.

Bases: ValueError.

This type’s implementation and inherited behavior are available in the linked release source.

Release source · Runtime guarantees · Capability guides

`EvalBaselineKey`

from cayu.evals import EvalBaselineKey

Stable scope of one explicit baseline pointer.

Bases: _EvalStoreModel.

Fields and attributes

Name	Type	Default
`target_key`	`StrictStr`	`required`
`corpus_revision`	`StrictStr`	`required`
`suite_id`	`StrictStr`	`required`

validate_portable_ids (method)

EvalBaselineKey.validate_portable_ids(value: str, info) -> str

validate_corpus_revision (method)

EvalBaselineKey.validate_corpus_revision(value: str, info) -> str

Release source · Runtime guarantees · Capability guides

`EvalBaselineMutationRecord`

from cayu.evals import EvalBaselineMutationRecord

Immutable audit fact for one committed baseline transition.

Bases: _EvalStoreModel.

Fields and attributes

Name	Type	Default
`operation_id`	`StrictStr`	`required`
`key`	`EvalBaselineKey`	`required`
`expected_generation`	`StrictInt`	`Field(ge=0, le=_EVAL_STORE_MAX_BIGINT)`
`previous_result_revision`	`StrictStr \| None`	`None`
`selected_result_revision`	`StrictStr`	`required`
`resulting_generation`	`StrictInt`	`Field(ge=1, le=_EVAL_STORE_MAX_BIGINT)`
`actor_id`	`StrictStr`	`required`
`created_at`	`datetime`	`required`

validate_revision_fields (method)

EvalBaselineMutationRecord.validate_revision_fields(value: str | None, info) -> str | None

validate_actor_id (method)

EvalBaselineMutationRecord.validate_actor_id(value: str, info) -> str

validate_created_at (method)

EvalBaselineMutationRecord.validate_created_at(value: datetime, info) -> datetime

validate_generations (method)

EvalBaselineMutationRecord.validate_generations() -> EvalBaselineMutationRecord

Release source · Runtime guarantees · Capability guides

`EvalBaselineRecord`

from cayu.evals import EvalBaselineRecord

Current mutable pointer to an immutable eval result.

Bases: _EvalStoreModel.

Fields and attributes

Name	Type	Default
`key`	`EvalBaselineKey`	`required`
`result_revision`	`StrictStr`	`required`
`generation`	`StrictInt`	`Field(ge=1, le=_EVAL_STORE_MAX_BIGINT)`
`updated_by`	`StrictStr`	`required`
`updated_at`	`datetime`	`required`

validate_result_revision (method)

EvalBaselineRecord.validate_result_revision(value: str, info) -> str

validate_updated_by (method)

EvalBaselineRecord.validate_updated_by(value: str, info) -> str

validate_updated_at (method)

EvalBaselineRecord.validate_updated_at(value: datetime, info) -> datetime

Release source · Runtime guarantees · Capability guides

`EvalBaselineUpdate`

from cayu.evals import EvalBaselineUpdate

Actor-attributed idempotent compare-and-swap baseline request.

Bases: _EvalStoreModel.

Fields and attributes

Name	Type	Default
`key`	`EvalBaselineKey`	`required`
`result_revision`	`StrictStr`	`required`
`expected_generation`	`StrictInt`	`Field(ge=0, le=_EVAL_STORE_MAX_BIGINT)`
`operation_id`	`StrictStr`	`required`
`actor_id`	`StrictStr`	`required`

validate_revision_fields (method)

EvalBaselineUpdate.validate_revision_fields(value: str, info) -> str

validate_actor_id (method)

EvalBaselineUpdate.validate_actor_id(value: str, info) -> str

Release source · Runtime guarantees · Capability guides

`EvalCaseCatalogEntry`

from cayu.evals import EvalCaseCatalogEntry

Bases: _EvalStoreModel.

Fields and attributes

Name	Type	Default
`corpus_revision`	`StrictStr`	`required`
`id`	`StrictStr`	`required`
`revision`	`StrictStr`	`required`
`suite_id`	`StrictStr`	`required`
`name`	`StrictStr`	`required`
`description`	`StrictStr \| None`	`None`
`message_count`	`StrictInt`	`Field(ge=0, le=EVAL_CORPUS_MAX_MESSAGES_PER_CASE)`
`assertion_count`	`StrictInt`	`Field(ge=1, le=EVAL_CORPUS_MAX_ASSERTIONS_PER_CASE)`

validate_revision_fields (method)

EvalCaseCatalogEntry.validate_revision_fields(value: str, info) -> str

validate_ids (method)

EvalCaseCatalogEntry.validate_ids(value: str, info) -> str

validate_name (method)

EvalCaseCatalogEntry.validate_name(value: str, info) -> str

validate_description (method)

EvalCaseCatalogEntry.validate_description(value: str | None, info) -> str | None

Release source · Runtime guarantees · Capability guides

`EvalCaseCatalogPage`

from cayu.evals import EvalCaseCatalogPage

Bases: _EvalStoreModel.

Fields and attributes

Name	Type	Default
`items`	`tuple[EvalCaseCatalogEntry, ...]`	`Field(max_length=EVAL_STORE_MAX_PAGE_SIZE)`
`next_cursor`	`StrictStr \| None`	`None`
`has_more`	`StrictBool`	`False`

validate_page (method)

EvalCaseCatalogPage.validate_page() -> EvalCaseCatalogPage

Release source · Runtime guarantees · Capability guides

`EvalCaseCatalogQuery`

from cayu.evals import EvalCaseCatalogQuery

Bases: _EvalStoreModel.

Fields and attributes

Name	Type	Default
`corpus_revision`	`StrictStr`	`required`
`suite_id`	`StrictStr`	`required`
`cursor`	`StrictStr \| None`	`None`
`limit`	`StrictInt`	`Field(default=EVAL_STORE_DEFAULT_PAGE_SIZE, ge=1, le=EVAL_STORE_MAX_PAGE_SIZE)`
`max_result_bytes`	`StrictInt`	`Field(default=EVAL_STORE_DEFAULT_PAGE_BYTES, ge=1024, le=EVAL_STORE_MAX_PAGE_BYTES)`

validate_revision (method)

EvalCaseCatalogQuery.validate_revision(value: str, info) -> str

validate_suite_id (method)

EvalCaseCatalogQuery.validate_suite_id(value: str, info) -> str

validate_cursor (method)

EvalCaseCatalogQuery.validate_cursor(value: str | None, info) -> str | None

Release source · Runtime guarantees · Capability guides

`EvalCatalogQuery`

from cayu.evals import EvalCatalogQuery

Bases: _EvalStoreModel.

Fields and attributes

Name	Type	Default
`target_key`	`StrictStr \| None`	`None`
`cursor`	`StrictStr \| None`	`None`
`limit`	`StrictInt`	`Field(default=EVAL_STORE_DEFAULT_PAGE_SIZE, ge=1, le=EVAL_STORE_MAX_PAGE_SIZE)`
`max_result_bytes`	`StrictInt`	`Field(default=EVAL_STORE_DEFAULT_PAGE_BYTES, ge=1024, le=EVAL_STORE_MAX_PAGE_BYTES)`

validate_target_key (method)

EvalCatalogQuery.validate_target_key(value: str | None, info) -> str | None

validate_cursor (method)

EvalCatalogQuery.validate_cursor(value: str | None, info) -> str | None

Release source · Runtime guarantees · Capability guides

`EvalCorpusCatalogEntry`

from cayu.evals import EvalCorpusCatalogEntry

Bases: _EvalStoreModel.

Fields and attributes

Name	Type	Default
`revision`	`StrictStr`	`required`
`target_key`	`StrictStr`	`required`
`evidence_policy_revision`	`StrictStr`	`required`
`pricing_profile_fingerprint`	`StrictStr \| None`	`None`
`suite_count`	`StrictInt`	`Field(ge=1, le=EVAL_CORPUS_MAX_SUITES)`
`case_count`	`StrictInt`	`Field(ge=1, le=EVAL_CORPUS_MAX_CASES)`
`assertion_count`	`StrictInt`	`Field(ge=1, le=EVAL_CORPUS_MAX_CASES * EVAL_CORPUS_MAX_ASSERTIONS_PER_CASE)`
`expanded_assertion_result_count`	`StrictInt`	`Field(ge=1, le=EVAL_CORPUS_MAX_SUITES * EVAL_CORPUS_MAX_PUBLISHED_ASSERTION_RESULTS)`
`document_bytes`	`StrictInt`	`Field(ge=1, le=EVAL_CORPUS_MAX_BYTES)`
`created_at`	`datetime`	`required`

validate_revision_fields (method)

EvalCorpusCatalogEntry.validate_revision_fields(value: str | None, info) -> str | None

validate_target_key (method)

EvalCorpusCatalogEntry.validate_target_key(value: str, info) -> str

validate_created_at (method)

EvalCorpusCatalogEntry.validate_created_at(value: datetime, info) -> datetime

validate_counts (method)

EvalCorpusCatalogEntry.validate_counts() -> EvalCorpusCatalogEntry

Release source · Runtime guarantees · Capability guides

`EvalCorpusCatalogPage`

from cayu.evals import EvalCorpusCatalogPage

Bases: _EvalStoreModel.

Fields and attributes

Name	Type	Default
`items`	`tuple[EvalCorpusCatalogEntry, ...]`	`Field(max_length=EVAL_STORE_MAX_PAGE_SIZE)`
`next_cursor`	`StrictStr \| None`	`None`
`has_more`	`StrictBool`	`False`

validate_page (method)

EvalCorpusCatalogPage.validate_page() -> EvalCorpusCatalogPage

Release source · Runtime guarantees · Capability guides

`EvalCorpusConflict`

from cayu.evals import EvalCorpusConflict

One immutable corpus revision resolves to contradictory stored content.

Bases: ValueError.

This type’s implementation and inherited behavior are available in the linked release source.

Release source · Runtime guarantees · Capability guides

`EvalJudgeCalibrationConflict`

from cayu.evals import EvalJudgeCalibrationConflict

A calibration revision or run identity resolves to contradictory content.

Bases: ValueError.

This type’s implementation and inherited behavior are available in the linked release source.

Release source · Runtime guarantees · Capability guides

`EvalResultConflict`

from cayu.evals import EvalResultConflict

One immutable result revision resolves to contradictory stored content.

Bases: ValueError.

This type’s implementation and inherited behavior are available in the linked release source.

Release source · Runtime guarantees · Capability guides

`EvalResultPage`

from cayu.evals import EvalResultPage

Bases: _EvalStoreModel.

Fields and attributes

Name	Type	Default
`items`	`tuple[EvalResultRecord, ...]`	`Field(max_length=EVAL_STORE_MAX_PAGE_SIZE)`
`next_cursor`	`StrictStr \| None`	`None`
`has_more`	`StrictBool`	`False`

validate_page (method)

EvalResultPage.validate_page() -> EvalResultPage

Release source · Runtime guarantees · Capability guides

`EvalResultQuery`

from cayu.evals import EvalResultQuery

Bounded target-scoped immutable result catalog query.

Bases: _EvalStoreModel.

Fields and attributes

Name	Type	Default
`target_key`	`StrictStr`	`required`
`origin`	`EvalResultOrigin \| None`	`None`
`cursor`	`StrictStr \| None`	`None`
`limit`	`StrictInt`	`Field(default=EVAL_STORE_DEFAULT_PAGE_SIZE, ge=1, le=EVAL_STORE_MAX_PAGE_SIZE)`
`max_result_bytes`	`StrictInt`	`Field(default=EVAL_STORE_DEFAULT_PAGE_BYTES, ge=1024, le=EVAL_STORE_MAX_PAGE_BYTES)`

validate_target_key (method)

EvalResultQuery.validate_target_key(value: str, info) -> str

validate_cursor (method)

EvalResultQuery.validate_cursor(value: str | None, info) -> str | None

Release source · Runtime guarantees · Capability guides

`EvalResultRecord`

from cayu.evals import EvalResultRecord

Bounded catalog metadata for one immutable captured or fresh result.

Bases: _EvalStoreModel.

Fields and attributes

Name	Type	Default
`revision`	`StrictStr`	`required`
`origin`	`EvalResultOrigin`	`required`
`target`	`EvalResultTargetIdentityV1`	`required`
`corpus_revision`	`StrictStr`	`required`
`suite_id`	`StrictStr`	`required`
`suite_revision`	`StrictStr`	`required`
`status`	`Literal['passed', 'failed', 'unavailable', 'error']`	`required`
`score`	`StrictFloat \| None`	`Field(default=None, ge=0.0, le=1.0)`
`document_bytes`	`StrictInt`	`Field(ge=1, le=CORPUS_EXECUTION_RESULT_MAX_BYTES)`
`created_at`	`datetime`	`required`

validate_revisions (method)

EvalResultRecord.validate_revisions(value: str, info) -> str

validate_suite_id (method)

EvalResultRecord.validate_suite_id(value: str, info) -> str

validate_created_at (method)

EvalResultRecord.validate_created_at(value: datetime, info) -> datetime

validate_result_summary (method)

EvalResultRecord.validate_result_summary() -> EvalResultRecord

Release source · Runtime guarantees · Capability guides

`EvalRunAdmissionConflict`

from cayu.evals import EvalRunAdmissionConflict

A run id or idempotency key is already bound to another request.

Bases: ValueError.

This type’s implementation and inherited behavior are available in the linked release source.

Release source · Runtime guarantees · Capability guides

`EvalRunClaim`

from cayu.evals import EvalRunClaim

Bases: _EvalStoreModel.

Fields and attributes

Name	Type	Default
`run_id`	`StrictStr`	`required`
`claim_id`	`StrictStr`	`Field(repr=False)`
`epoch`	`StrictInt`	`Field(ge=1, le=_EVAL_STORE_MAX_BIGINT)`

validate_ids (method)

EvalRunClaim.validate_ids(value: str, info) -> str

Release source · Runtime guarantees · Capability guides

`EvalRunClaimLost`

from cayu.evals import EvalRunClaimLost

A worker no longer owns the live fenced claim required for a mutation.

Bases: RuntimeError.

This type’s implementation and inherited behavior are available in the linked release source.

Release source · Runtime guarantees · Capability guides

`EvalRunCostBudget`

from cayu.evals import EvalRunCostBudget

One server-priced cost ceiling applied independently to each eval-trial session.

Bases: _EvalStoreModel.

Fields and attributes

Name	Type	Default
`max_estimated_cost`	`Decimal`	`Field(gt=0)`
`currency`	`StrictStr`	`Field(default='USD', min_length=1, max_length=16)`

validate_cost (method)

EvalRunCostBudget.validate_cost(value: Decimal, info) -> Decimal

validate_currency (method)

EvalRunCostBudget.validate_currency(value: str, info) -> str

Release source · Runtime guarantees · Capability guides

`EvalRunFailureCode`

from cayu.evals import EvalRunFailureCode

Safe terminal diagnostics; arbitrary exception text is never persisted.

Bases: StrEnum.

Fields and attributes

Name	Type	Default
`TARGET_UNAVAILABLE`	`class attribute`	`'target_unavailable'`
`CORPUS_UNAVAILABLE`	`class attribute`	`'corpus_unavailable'`
`EXECUTION_FAILED`	`class attribute`	`'execution_failed'`
`WORKER_INTERRUPTED`	`class attribute`	`'worker_interrupted'`

Release source · Runtime guarantees · Capability guides

`EvalRunFailureDiagnostic`

from cayu.evals import EvalRunFailureDiagnostic

Safe causal evidence bound to the containing run and fenced terminal claim.

Bases: _EvalStoreModel.

Fields and attributes

Name	Type	Default
`reason`	`EvalRunFailureReason`	`required`
`provider_protocol_reason`	`StrictStr \| None`	`Field(default=None, max_length=128)`

validate_protocol_reason (method)

EvalRunFailureDiagnostic.validate_protocol_reason(value: str | None) -> str | None

validate_protocol_failure (method)

EvalRunFailureDiagnostic.validate_protocol_failure() -> EvalRunFailureDiagnostic

phase (property)

EvalRunFailureDiagnostic.phase() -> str

Release source · Runtime guarantees · Capability guides

`EvalRunFailureReason`

from cayu.evals import EvalRunFailureReason

Closed causal vocabulary; values never come from exception messages.

Bases: StrEnum.

Fields and attributes

Name	Type	Default
`CORPUS_LOAD_FAILED`	`class attribute`	`'corpus_load_failed'`
`CORPUS_MISSING`	`class attribute`	`'corpus_missing'`
`TARGET_REGISTRATION_MISSING`	`class attribute`	`'target_registration_missing'`
`TARGET_PREFLIGHT_FAILED`	`class attribute`	`'target_preflight_failed'`
`EXECUTION_PROFILE_MISSING`	`class attribute`	`'execution_profile_missing'`
`EXECUTION_PROFILE_CHANGED`	`class attribute`	`'execution_profile_changed'`
`TARGET_IDENTITY_FAILED`	`class attribute`	`'target_identity_failed'`
`TARGET_IDENTITY_CHANGED`	`class attribute`	`'target_identity_changed'`
`CORPUS_COMPILATION_FAILED`	`class attribute`	`'corpus_compilation_failed'`
`EXECUTION_FAILED`	`class attribute`	`'execution_failed'`
`EXECUTION_PROFILE_REJECTED`	`class attribute`	`'execution_profile_rejected'`
`PROVIDER_PROTOCOL_FAILED`	`class attribute`	`'provider_protocol_failed'`
`EXECUTION_CANCELLED`	`class attribute`	`'execution_cancelled'`
`RESULT_PUBLICATION_FAILED`	`class attribute`	`'result_publication_failed'`

Release source · Runtime guarantees · Capability guides

`EvalRunInvocation`

from cayu.evals import EvalRunInvocation

Durable, authority-free execution contractions and trusted caller provenance.

The HTTP API constructs this value after authentication. It retains only the bounded subject/tenant projection needed to mint ordinary runtime invocation provenance after a worker restart; authentication claims never enter EvalStore. None bounds inherit the server-owned target request base.

Bases: _EvalStoreModel.

Fields and attributes

Name	Type	Default
`schema_version`	`Literal[1]`	`1`
`retain_trial_checkpoints`	`StrictBool`	`Field(default=False, exclude_if=lambda value: value is False)`
`recovery_policy`	`EvalRunRecoveryPolicyV1 \| None`	`Field(default=None, exclude_if=lambda value: value is None)`
`retry_of`	`EvalRunRetryLineageV1 \| None`	`Field(default=None, exclude_if=lambda value: value is None)`
`source`	`SessionExecutionSource`	`SessionExecutionSource.SDK_RUN`
`origin`	`InvocationOrigin \| None`	`None`
`max_steps`	`StrictInt \| None`	`Field(default=None, ge=1, le=MAX_STEPS)`
`limits`	`RunLimits \| None`	`None`
`cost_budget`	`EvalRunCostBudget \| None`	`None`
`execution_profile`	`EvalExecutionProfileBindingV1 \| None`	`Field(default=None, exclude_if=lambda value: value is None)`
`execution_profile_snapshot`	`EvalExecutionProfileV1 \| None`	`Field(default=None, exclude_if=lambda value: value is None)`
`admission_request_revision`	`StrictStr \| None`	`Field(default=None, description='Secret-safe identity of the exact authenticated request that admitted this run.', exclude_if=lambda value: value is None)`
`authored_suite_revision`	`StrictStr \| None`	`Field(default=None, exclude_if=lambda value: value is None)`
`authored_suite_selection_revision`	`StrictStr \| None`	`Field(default=None, exclude_if=lambda value: value is None)`
`authored_suite_launch_revision`	`StrictStr \| None`	`Field(default=None, description='Secret-safe identity shared by every durable run admitted by one authored-suite launch.', exclude_if=lambda value: value is None)`
`authored_suite_launch_lane`	`StrictInt \| None`	`Field(default=None, ge=0, le=EVAL_SUITE_MAX_CONCURRENCY - 1, description='Concurrency lane shared by serialized runs from one suite launch.', exclude_if=lambda value: value is None)`
`authored_suite_exposure`	`EvalSuiteRunExposureV1 \| None`	`Field(default=None, exclude_if=lambda value: value is None)`
`scenario`	`EvalScenarioRunInvocation \| None`	`Field(default=None, exclude_if=lambda value: value is None)`

validate_schema_version_type (method)

EvalRunInvocation.validate_schema_version_type(value: object) -> object

copy_origin (method)

EvalRunInvocation.copy_origin(value: object) -> object

copy_limits (method)

EvalRunInvocation.copy_limits(value: object) -> object

copy_cost_budget (method)

EvalRunInvocation.copy_cost_budget(value: object) -> object

copy_execution_profile (method)

EvalRunInvocation.copy_execution_profile(value: object) -> object

copy_execution_profile_snapshot (method)

EvalRunInvocation.copy_execution_profile_snapshot(value: object) -> object

copy_scenario (method)

EvalRunInvocation.copy_scenario(value: object) -> object

copy_authored_suite_exposure (method)

EvalRunInvocation.copy_authored_suite_exposure(value: object) -> object

validate_content_revisions (method)

EvalRunInvocation.validate_content_revisions(value: str | None, info) -> str | None

validate_provenance (method)

EvalRunInvocation.validate_provenance() -> EvalRunInvocation

Release source · Runtime guarantees · Capability guides

`EvalRunLease`

from cayu.evals import EvalRunLease

Bases: _EvalStoreModel.

Fields and attributes

Name	Type	Default
`run`	`EvalRunRecord`	`required`
`claim`	`EvalRunClaim`	`Field(repr=False)`

validate_binding (method)

EvalRunLease.validate_binding() -> EvalRunLease

Release source · Runtime guarantees · Capability guides

`EvalRunObservation`

from cayu.evals import EvalRunObservation

Small mutable-lifecycle projection that never rehydrates run invocation data.

Bases: _EvalStoreModel.

Fields and attributes

Name	Type	Default
`run_id`	`StrictStr`	`required`
`status`	`EvalRunStatus`	`required`
`attempt_count`	`StrictInt`	`Field(ge=0, le=_EVAL_STORE_MAX_BIGINT)`
`updated_at`	`datetime`	`required`
`ownership`	`EvalRunOwnership \| None`	`None`

validate_run_id (method)

EvalRunObservation.validate_run_id(value: str, info) -> str

validate_updated_at (method)

EvalRunObservation.validate_updated_at(value: datetime, info) -> datetime

validate_lifecycle (method)

EvalRunObservation.validate_lifecycle() -> EvalRunObservation

id (property)

EvalRunObservation.id() -> str

terminal (property)

EvalRunObservation.terminal() -> bool

Release source · Runtime guarantees · Capability guides

`EvalRunOwnership`

from cayu.evals import EvalRunOwnership

Bases: _EvalStoreModel.

Fields and attributes

Name	Type	Default
`epoch`	`StrictInt`	`Field(ge=1, le=_EVAL_STORE_MAX_BIGINT)`
`lease_expires_at`	`datetime`	`required`

validate_lease_expires_at (method)

EvalRunOwnership.validate_lease_expires_at(value: datetime, info) -> datetime

Release source · Runtime guarantees · Capability guides

`EvalRunPage`

from cayu.evals import EvalRunPage

Bases: _EvalStoreModel.

Fields and attributes

Name	Type	Default
`items`	`tuple[EvalRunRecord, ...]`	`Field(max_length=EVAL_STORE_MAX_PAGE_SIZE)`
`next_cursor`	`StrictStr \| None`	`None`
`has_more`	`StrictBool`	`False`

validate_page (method)

EvalRunPage.validate_page() -> EvalRunPage

Release source · Runtime guarantees · Capability guides

`EvalRunQuery`

from cayu.evals import EvalRunQuery

Bases: _EvalStoreModel.

Fields and attributes

Name	Type	Default
`target_key`	`StrictStr \| None`	`None`
`status`	`EvalRunStatus \| None`	`None`
`corpus_revision`	`StrictStr \| None`	`None`
`cursor`	`StrictStr \| None`	`None`
`limit`	`StrictInt`	`Field(default=EVAL_STORE_DEFAULT_PAGE_SIZE, ge=1, le=EVAL_STORE_MAX_PAGE_SIZE)`
`max_result_bytes`	`StrictInt`	`Field(default=EVAL_STORE_DEFAULT_PAGE_BYTES, ge=1024, le=EVAL_STORE_MAX_PAGE_BYTES)`

validate_target_key (method)

EvalRunQuery.validate_target_key(value: str | None, info) -> str | None

validate_revision (method)

EvalRunQuery.validate_revision(value: str | None, info) -> str | None

validate_cursor (method)

EvalRunQuery.validate_cursor(value: str | None, info) -> str | None

Release source · Runtime guarantees · Capability guides

`EvalRunRecord`

from cayu.evals import EvalRunRecord

Bases: _EvalStoreModel.

Fields and attributes

Name	Type	Default
`spec`	`EvalRunSpec`	`required`
`status`	`EvalRunStatus`	`required`
`attempt_count`	`StrictInt`	`Field(ge=0, le=_EVAL_STORE_MAX_BIGINT)`
`created_at`	`datetime`	`required`
`updated_at`	`datetime`	`required`
`started_at`	`datetime \| None`	`None`
`finished_at`	`datetime \| None`	`None`
`cancel_requested_at`	`datetime \| None`	`None`
`ownership`	`EvalRunOwnership \| None`	`None`
`result`	`EvalRunResultSummary \| None`	`None`
`failure_code`	`EvalRunFailureCode \| None`	`None`
`failure_diagnostic`	`EvalRunFailureDiagnostic \| None`	`None`
`scenario_progress`	`EvalScenarioRunProgress \| None`	`Field(default=None, exclude_if=lambda value: value is None)`

validate_timestamps (method)

EvalRunRecord.validate_timestamps(value: datetime | None, info) -> datetime | None

validate_lifecycle (method)

EvalRunRecord.validate_lifecycle() -> EvalRunRecord

id (property)

EvalRunRecord.id() -> str

Release source · Runtime guarantees · Capability guides

`EvalRunRequest`

from cayu.evals import EvalRunRequest

Bases: EvalRunSpec.

Fields and attributes

Name	Type	Default
`run_id`	`StrictStr`	`required`
`corpus_revision`	`StrictStr`	`required`
`target_key`	`StrictStr`	`required`
`suite_id`	`StrictStr`	`required`
`suite_revision`	`StrictStr`	`required`
`max_concurrency`	`StrictInt`	`Field(ge=1, le=EVAL_MAX_CONCURRENCY)`
`invocation`	`EvalRunInvocation`	`Field(default_factory=EvalRunInvocation)`
`idempotency_key`	`StrictStr`	`Field(repr=False)`

validate_run_id (method)

EvalRunRequest.validate_run_id(value: str, info) -> str

validate_revisions (method)

EvalRunRequest.validate_revisions(value: str, info) -> str

validate_portable_ids (method)

EvalRunRequest.validate_portable_ids(value: str, info) -> str

copy_invocation (method)

EvalRunRequest.copy_invocation(value: object) -> object

id (property)

EvalRunRequest.id() -> str

validate_idempotency_key (method)

EvalRunRequest.validate_idempotency_key(value: str, info) -> str

same_logical_request (method)

EvalRunRequest.same_logical_request(other: EvalRunRequest) -> bool

Release source · Runtime guarantees · Capability guides

`EvalRunResultSummary`

from cayu.evals import EvalRunResultSummary

Bases: _EvalStoreModel.

Fields and attributes

Name	Type	Default
`revision`	`StrictStr`	`required`
`status`	`Literal['passed', 'failed', 'unavailable', 'error']`	`required`
`score`	`StrictFloat \| None`	`Field(default=None, ge=0.0, le=1.0)`
`duration_ms`	`StrictInt`	`Field(ge=0, le=_EVAL_STORE_MAX_BIGINT)`

validate_revision (method)

EvalRunResultSummary.validate_revision(value: str, info) -> str

validate_score (method)

EvalRunResultSummary.validate_score() -> EvalRunResultSummary

Release source · Runtime guarantees · Capability guides

`EvalRunSpec`

from cayu.evals import EvalRunSpec

Bases: _EvalStoreModel.

Fields and attributes

Name	Type	Default
`run_id`	`StrictStr`	`required`
`corpus_revision`	`StrictStr`	`required`
`target_key`	`StrictStr`	`required`
`suite_id`	`StrictStr`	`required`
`suite_revision`	`StrictStr`	`required`
`max_concurrency`	`StrictInt`	`Field(ge=1, le=EVAL_MAX_CONCURRENCY)`
`invocation`	`EvalRunInvocation`	`Field(default_factory=EvalRunInvocation)`

validate_run_id (method)

EvalRunSpec.validate_run_id(value: str, info) -> str

validate_revisions (method)

EvalRunSpec.validate_revisions(value: str, info) -> str

validate_portable_ids (method)

EvalRunSpec.validate_portable_ids(value: str, info) -> str

copy_invocation (method)

EvalRunSpec.copy_invocation(value: object) -> object

id (property)

EvalRunSpec.id() -> str

Release source · Runtime guarantees · Capability guides

`EvalRunStateConflict`

from cayu.evals import EvalRunStateConflict

A requested run lifecycle transition is not valid from current state.

Bases: ValueError.

This type’s implementation and inherited behavior are available in the linked release source.

Release source · Runtime guarantees · Capability guides

`EvalRunStatus`

from cayu.evals import EvalRunStatus

Bases: StrEnum.

Fields and attributes

Name	Type	Default
`QUEUED`	`class attribute`	`'queued'`
`RUNNING`	`class attribute`	`'running'`
`CANCELLING`	`class attribute`	`'cancelling'`
`COMPLETED`	`class attribute`	`'completed'`
`FAILED`	`class attribute`	`'failed'`
`CANCELLED`	`class attribute`	`'cancelled'`

Release source · Runtime guarantees · Capability guides

`EvalScenarioApprovalDecisionRecord`

from cayu.evals import EvalScenarioApprovalDecisionRecord

Bases: _EvalStoreModel.

Fields and attributes

Name	Type	Default
`decision`	`Literal['approve', 'deny']`	`required`
`reason`	`StrictStr \| None`	`Field(default=None, max_length=2048)`
`actor_id`	`StrictStr`	`Field(max_length=512)`
`submitted_at`	`datetime`	`required`

validate_reason (method)

EvalScenarioApprovalDecisionRecord.validate_reason(value: str | None, info) -> str | None

validate_actor_id (method)

EvalScenarioApprovalDecisionRecord.validate_actor_id(value: str, info) -> str

validate_submitted_at (method)

EvalScenarioApprovalDecisionRecord.validate_submitted_at(value: datetime, info) -> datetime

Release source · Runtime guarantees · Capability guides

`EvalScenarioApprovalSubmission`

from cayu.evals import EvalScenarioApprovalSubmission

Bases: _EvalStoreModel.

Fields and attributes

Name	Type	Default
`expected_progress_revision`	`StrictStr`	`required`
`trial_number`	`StrictInt`	`Field(ge=1, le=EVAL_CORPUS_MAX_TRIALS)`
`event_id`	`StrictStr`	`required`
`decision`	`Literal['approve', 'deny']`	`required`
`reason`	`StrictStr \| None`	`Field(default=None, max_length=2048)`
`actor_id`	`StrictStr`	`Field(max_length=512)`

validate_expected_revision (method)

EvalScenarioApprovalSubmission.validate_expected_revision(value: str, info) -> str

validate_event_id (method)

EvalScenarioApprovalSubmission.validate_event_id(value: str, info) -> str

validate_reason (method)

EvalScenarioApprovalSubmission.validate_reason(value: str | None, info) -> str | None

validate_actor_id (method)

EvalScenarioApprovalSubmission.validate_actor_id(value: str, info) -> str

Release source · Runtime guarantees · Capability guides

`EvalScenarioArtifactReference`

from cayu.evals import EvalScenarioArtifactReference

One immutable scenario requirement-to-artifact launch selection.

Bases: _EvalStoreModel.

Fields and attributes

Name	Type	Default
`requirement_id`	`StrictStr`	`required`
`artifact_id`	`StrictStr`	`required`

validate_requirement_id (method)

EvalScenarioArtifactReference.validate_requirement_id(value: str, info) -> str

validate_artifact_id (method)

EvalScenarioArtifactReference.validate_artifact_id(value: str, info) -> str

Release source · Runtime guarantees · Capability guides

`EvalScenarioCatalogEntry`

from cayu.evals import EvalScenarioCatalogEntry

Bases: _EvalStoreModel.

Fields and attributes

Name	Type	Default
`revision`	`StrictStr`	`required`
`id`	`StrictStr`	`required`
`target_key`	`StrictStr`	`required`
`name`	`StrictStr`	`required`
`description`	`StrictStr \| None`	`None`
`event_count`	`StrictInt`	`Field(ge=1, le=EVAL_SCENARIO_MAX_EVENTS)`
`input_event_count`	`StrictInt`	`Field(ge=1, le=EVAL_SCENARIO_MAX_EVENTS)`
`approval_checkpoint_count`	`StrictInt`	`Field(ge=0, le=EVAL_SCENARIO_MAX_EVENTS)`
`message_count`	`StrictInt`	`Field(ge=1, le=EVAL_SCENARIO_MAX_EVENTS * 32)`
`part_count`	`StrictInt`	`Field(ge=1, le=EVAL_SCENARIO_MAX_EVENTS * 32 * 32)`
`artifact_requirement_count`	`StrictInt`	`Field(ge=0, le=EVAL_SCENARIO_MAX_ARTIFACT_REQUIREMENTS)`
`secret_requirement_count`	`StrictInt`	`Field(ge=0, le=EVAL_SCENARIO_MAX_SECRET_REQUIREMENTS)`
`document_bytes`	`StrictInt`	`Field(ge=1, le=EVAL_SCENARIO_MAX_BYTES)`
`created_at`	`datetime`	`required`

validate_revision (method)

EvalScenarioCatalogEntry.validate_revision(value: str, info) -> str

validate_ids (method)

EvalScenarioCatalogEntry.validate_ids(value: str, info) -> str

validate_name (method)

EvalScenarioCatalogEntry.validate_name(value: str, info) -> str

validate_description (method)

EvalScenarioCatalogEntry.validate_description(value: str | None, info) -> str | None

validate_created_at (method)

EvalScenarioCatalogEntry.validate_created_at(value: datetime, info) -> datetime

validate_counts (method)

EvalScenarioCatalogEntry.validate_counts() -> EvalScenarioCatalogEntry

Release source · Runtime guarantees · Capability guides

`EvalScenarioCatalogPage`

from cayu.evals import EvalScenarioCatalogPage

Bases: _EvalStoreModel.

Fields and attributes

Name	Type	Default
`items`	`tuple[EvalScenarioCatalogEntry, ...]`	`Field(max_length=EVAL_STORE_MAX_PAGE_SIZE)`
`next_cursor`	`StrictStr \| None`	`None`
`has_more`	`StrictBool`	`False`

validate_page (method)

EvalScenarioCatalogPage.validate_page() -> EvalScenarioCatalogPage

Release source · Runtime guarantees · Capability guides

`EvalScenarioCatalogQuery`

from cayu.evals import EvalScenarioCatalogQuery

Bases: _EvalStoreModel.

Fields and attributes

Name	Type	Default
`target_key`	`StrictStr \| None`	`None`
`scenario_id`	`StrictStr \| None`	`None`
`cursor`	`StrictStr \| None`	`None`
`limit`	`StrictInt`	`Field(default=EVAL_STORE_DEFAULT_PAGE_SIZE, ge=1, le=EVAL_STORE_MAX_PAGE_SIZE)`
`max_result_bytes`	`StrictInt`	`Field(default=EVAL_STORE_DEFAULT_PAGE_BYTES, ge=1024, le=EVAL_STORE_MAX_PAGE_BYTES)`

validate_ids (method)

EvalScenarioCatalogQuery.validate_ids(value: str | None, info) -> str | None

validate_cursor (method)

EvalScenarioCatalogQuery.validate_cursor(value: str | None, info) -> str | None

Release source · Runtime guarantees · Capability guides

`EvalScenarioConflict`

from cayu.evals import EvalScenarioConflict

One immutable scenario revision resolves to contradictory stored content.

Bases: ValueError.

This type’s implementation and inherited behavior are available in the linked release source.

Release source · Runtime guarantees · Capability guides

`EvalScenarioRunInvocation`

from cayu.evals import EvalScenarioRunInvocation

Authority-free scenario launch facts frozen into one durable run.

Bases: _EvalStoreModel.

Fields and attributes

Name	Type	Default
`schema_version`	`Literal[1]`	`1`
`scenario_revision`	`StrictStr`	`required`
`binding_revision`	`StrictStr`	`required`
`authored_suite_revision`	`StrictStr \| None`	`Field(default=None, exclude_if=lambda value: value is None)`
`authored_case_revision`	`StrictStr \| None`	`Field(default=None, exclude_if=lambda value: value is None)`
`environment_name`	`StrictStr \| None`	`None`
`trials`	`StrictInt`	`Field(ge=1, le=EVAL_CORPUS_MAX_TRIALS)`
`timeout_seconds`	`StrictInt`	`Field(ge=1, le=EVAL_CORPUS_MAX_TIMEOUT_SECONDS)`
`artifact_references`	`tuple[EvalScenarioArtifactReference, ...]`	`Field(default_factory=tuple, max_length=EVAL_SCENARIO_MAX_ARTIFACT_REQUIREMENTS)`

validate_schema_version_type (method)

EvalScenarioRunInvocation.validate_schema_version_type(value: object) -> object

validate_revisions (method)

EvalScenarioRunInvocation.validate_revisions(value: str | None, info) -> str | None

validate_environment_name (method)

EvalScenarioRunInvocation.validate_environment_name(value: str | None, info) -> str | None

validate_artifact_references (method)

EvalScenarioRunInvocation.validate_artifact_references() -> EvalScenarioRunInvocation

Release source · Runtime guarantees · Capability guides

`EvalScenarioRunProgress`

from cayu.evals import EvalScenarioRunProgress

Bases: _EvalStoreModel.

Fields and attributes

Name	Type	Default
`schema_version`	`Literal[1]`	`1`
`revision`	`StrictStr`	`required`
`scenario_revision`	`StrictStr`	`required`
`binding_revision`	`StrictStr`	`required`
`attempt`	`StrictInt`	`Field(ge=1, le=_EVAL_STORE_MAX_BIGINT)`
`trials`	`tuple[EvalScenarioTrialProgress, ...]`	`Field(min_length=1, max_length=EVAL_CORPUS_MAX_TRIALS)`

validate_schema_version_type (method)

EvalScenarioRunProgress.validate_schema_version_type(value: object) -> object

validate_revisions (method)

EvalScenarioRunProgress.validate_revisions(value: str, info) -> str

validate_progress (method)

EvalScenarioRunProgress.validate_progress() -> EvalScenarioRunProgress

create (method)

EvalScenarioRunProgress.create(*, scenario_revision: str, binding_revision: str, attempt: int, trials: tuple[EvalScenarioTrialProgress, ...]) -> EvalScenarioRunProgress

replace_trial (method)

EvalScenarioRunProgress.replace_trial(trial: EvalScenarioTrialProgress) -> EvalScenarioRunProgress

Release source · Runtime guarantees · Capability guides

`EvalScenarioTrialFailureCode`

from cayu.evals import EvalScenarioTrialFailureCode

Bases: StrEnum.

Fields and attributes

Name	Type	Default
`EXECUTION_FAILED`	`class attribute`	`'execution_failed'`
`EXECUTION_PROFILE_CHANGED`	`class attribute`	`'execution_profile_changed'`
`UNEXPECTED_SESSION_STATE`	`class attribute`	`'unexpected_session_state'`
`EXPECTED_APPROVAL_UNAVAILABLE`	`class attribute`	`'expected_approval_unavailable'`
`EXPECTED_USER_INPUT_UNAVAILABLE`	`class attribute`	`'expected_user_input_unavailable'`

Release source · Runtime guarantees · Capability guides

`EvalScenarioTrialPhase`

from cayu.evals import EvalScenarioTrialPhase

Bases: StrEnum.

Fields and attributes

Name	Type	Default
`PENDING`	`class attribute`	`'pending'`
`RUNNING`	`class attribute`	`'running'`
`AWAITING_APPROVAL`	`class attribute`	`'awaiting_approval'`
`AWAITING_RESUME`	`class attribute`	`'awaiting_resume'`
`COMPLETED`	`class attribute`	`'completed'`
`ERROR`	`class attribute`	`'error'`

Release source · Runtime guarantees · Capability guides

`EvalScenarioTrialProgress`

from cayu.evals import EvalScenarioTrialProgress

Bases: _EvalStoreModel.

Fields and attributes

Name	Type	Default
`trial_number`	`StrictInt`	`Field(ge=1, le=EVAL_CORPUS_MAX_TRIALS)`
`phase`	`EvalScenarioTrialPhase`	`required`
`session_id`	`StrictStr \| None`	`None`
`next_event_sequence`	`StrictInt`	`Field(ge=0, le=EVAL_SCENARIO_MAX_EVENTS)`
`pending_event_id`	`StrictStr \| None`	`None`
`pending_tool_name`	`StrictStr \| None`	`None`
`pending_input_id`	`StrictStr \| None`	`None`
`pending_resume_kind`	`Literal['user_input', 'manual_recovery'] \| None`	`None`
`approval`	`EvalScenarioApprovalDecisionRecord \| None`	`None`
`failure_code`	`EvalScenarioTrialFailureCode \| None`	`None`

validate_session_id (method)

EvalScenarioTrialProgress.validate_session_id(value: str | None, info) -> str | None

validate_pending_event_id (method)

EvalScenarioTrialProgress.validate_pending_event_id(value: str | None, info) -> str | None

validate_pending_tool_name (method)

EvalScenarioTrialProgress.validate_pending_tool_name(value: str | None, info) -> str | None

validate_pending_input_id (method)

EvalScenarioTrialProgress.validate_pending_input_id(value: str | None, info) -> str | None

validate_phase (method)

EvalScenarioTrialProgress.validate_phase() -> EvalScenarioTrialProgress

Release source · Runtime guarantees · Capability guides

`EvalStore`

from cayu.evals import EvalStore

Bounded persistence for public eval corpora, run state, and safe results.

Bases: ABC.

close (method)

async EvalStore.close() -> None

Release store resources; process-local stores may perform no work.

save_corpus (method)

async EvalStore.save_corpus(corpus: EvalCorpusDocument, *, redact_json: Callable[[Any], Any]) -> EvalCorpusCatalogEntry

Scan and atomically save one complete immutable corpus revision.

load_corpus (method)

async EvalStore.load_corpus(revision: str, *, max_bytes: int=EVAL_CORPUS_MAX_BYTES) -> EvalCorpusDocument | None

Load one corpus without crossing the caller’s byte ceiling.

list_corpora (method)

async EvalStore.list_corpora(query: EvalCatalogQuery | None=None) -> EvalCorpusCatalogPage

List immutable corpus revisions in newest-first keyset order.

save_scenario (method)

async EvalStore.save_scenario(scenario: EvalScenarioDocumentV2, *, redact_json: Callable[[Any], Any]) -> EvalScenarioCatalogEntry

Scan and atomically save one immutable portable scenario revision.

load_scenario (method)

async EvalStore.load_scenario(revision: str, *, max_bytes: int=EVAL_SCENARIO_MAX_BYTES) -> EvalScenarioDocumentV2 | None

Load one scenario without crossing the caller’s byte ceiling.

list_scenarios (method)

async EvalStore.list_scenarios(query: EvalScenarioCatalogQuery | None=None) -> EvalScenarioCatalogPage

List immutable scenario revisions in newest-first keyset order.

save_authored_suite (method)

async EvalStore.save_authored_suite(document: EvalSuiteDocument, *, redact_json: Callable[[Any], Any]) -> EvalAuthoredSuiteCatalogEntry

Scan and atomically save one reviewed immutable authored suite.

load_authored_suite (method)

async EvalStore.load_authored_suite(revision: str, *, max_bytes: int=EVAL_SUITE_AUTHORING_MAX_BYTES) -> EvalSuiteDocument | None

Load one authored suite without crossing the caller byte ceiling.

list_authored_suites (method)

async EvalStore.list_authored_suites(query: EvalAuthoredSuiteCatalogQuery | None=None) -> EvalAuthoredSuiteCatalogPage

List immutable authored suite revisions in newest-first keyset order.

save_judge_calibration (method)

async EvalStore.save_judge_calibration(report: EvalJudgeCalibrationReportV1, *, redact_json: Callable[[Any], Any]) -> EvalJudgeCalibrationReportV1

Scan and atomically retain every trial in one completed calibration.

load_judge_calibration (method)

async EvalStore.load_judge_calibration(revision: str, *, max_bytes: int=EVAL_JUDGE_CALIBRATION_MAX_BYTES) -> EvalJudgeCalibrationReportV1 | None

Load one immutable completed calibration by content revision.

load_judge_calibration_by_run_id (method)

async EvalStore.load_judge_calibration_by_run_id(run_id: str, *, max_bytes: int=EVAL_JUDGE_CALIBRATION_MAX_BYTES) -> EvalJudgeCalibrationReportV1 | None

Recover one completed calibration after retry or server restart.

list_suites (method)

async EvalStore.list_suites(query: EvalSuiteCatalogQuery) -> EvalSuiteCatalogPage

List bounded suite projections for one corpus revision.

list_cases (method)

async EvalStore.list_cases(query: EvalCaseCatalogQuery) -> EvalCaseCatalogPage

List bounded case projections for one corpus suite.

admit_run (method)

async EvalStore.admit_run(request: EvalRunRequest, *, redact_json: Callable[[Any], Any]) -> EvalRunRecord

Atomically admit or replay one logical run request.

load_run_by_idempotency_key (method)

async EvalStore.load_run_by_idempotency_key(idempotency_key: str) -> EvalRunRecord | None

Load a previously admitted run by its server-scoped idempotency digest.

load_run (method)

async EvalStore.load_run(run_id: str) -> EvalRunRecord | None

Load one run’s public lifecycle projection.

load_run_observation (method)

async EvalStore.load_run_observation(run_id: str) -> EvalRunObservation | None

Load mutable claim/status state without requiring immutable invocation data.

Custom stores retain compatibility through this full-record fallback. Built-in stores override it with a small indexed projection.

wait_for_run_terminal (method)

async EvalStore.wait_for_run_terminal(run_id: str, *, timeout_seconds: float=30.0, poll_interval_seconds: float=EVAL_RUN_MIN_OBSERVATION_INTERVAL_SECONDS, max_poll_interval_seconds: float=EVAL_RUN_MAX_OBSERVATION_INTERVAL_SECONDS) -> EvalRunObservation | None

Wait within a strict deadline for a terminal run observation.

Polling begins no faster than the public 50 ms floor and exponentially backs off to the caller’s bounded ceiling. A missing run returns None; deadline exhaustion raises TimeoutError.

list_runs (method)

async EvalStore.list_runs(query: EvalRunQuery | None=None) -> EvalRunPage

List run projections in newest-first keyset order.

claim_run (method)

async EvalStore.claim_run(*, target_key: str | None=None, launch_revision: str | None=None, lease_seconds: int=300) -> EvalRunLease | None

Claim eligible work, optionally restricted to one exact authored launch.

claim_run_for_targets (method)

async EvalStore.claim_run_for_targets(target_keys: tuple[str, ...], *, lease_seconds: int=300) -> EvalRunLease | None

Claim work only from a bounded approved target set.

Custom stores inherit a compatible sequential fallback. Durable built-in stores override this with one indexed atomic claim across the whole set.

heartbeat_run (method)

async EvalStore.heartbeat_run(claim: EvalRunClaim, *, extend_seconds: int=300) -> EvalRunRecord

Extend one still-live fenced run claim.

heartbeat_run_observation (method)

async EvalStore.heartbeat_run_observation(claim: EvalRunClaim, *, extend_seconds: int=300) -> EvalRunObservation

Extend a claim and return only its mutable lifecycle projection.

Custom stores retain compatibility through the full-record heartbeat. Built-in stores override this method to avoid immutable rehydration.

request_cancel (method)

async EvalStore.request_cancel(run_id: str) -> EvalRunRecord

Persist cancellation intent, terminalizing unclaimed queued work.

initialize_scenario_progress (method)

async EvalStore.initialize_scenario_progress(claim: EvalRunClaim, progress: EvalScenarioRunProgress) -> EvalRunRecord

Replace scenario progress for a newly claimed execution attempt.

update_scenario_trial (method)

async EvalStore.update_scenario_trial(claim: EvalRunClaim, trial: EvalScenarioTrialProgress) -> EvalRunRecord

Fenced update of one trial inside the current scenario attempt.

load_trial_checkpoints (method)

async EvalStore.load_trial_checkpoints(claim: EvalRunClaim) -> tuple[EvalRunTrialCheckpoint, ...]

Load terminal trial slots only while the exact run claim remains live.

load_trial_evidence_links (method)

async EvalStore.load_trial_evidence_links(run_id: str) -> tuple[EvalTrialEvidenceLinkV1, ...]

Read private retained links without execution authority; unsupported stores return none.

save_trial_checkpoint (method)

async EvalStore.save_trial_checkpoint(claim: EvalRunClaim, checkpoint: EvalRunTrialCheckpoint, *, redact_json: Callable[[Any], Any]) -> None

Idempotently retain one terminal slot behind the live publication fence.

submit_scenario_approval (method)

async EvalStore.submit_scenario_approval(run_id: str, submission: EvalScenarioApprovalSubmission) -> EvalRunRecord

CAS one fresh operator decision into the exact pending checkpoint.

publish_result (method)

async EvalStore.publish_result(claim: EvalRunClaim, result: CorpusExecutionResult, *, redact_json: Callable[[Any], Any]) -> EvalRunRecord

Atomically publish exactly one immutable safe terminal result.

fail_run (method)

async EvalStore.fail_run(claim: EvalRunClaim, code: EvalRunFailureCode, *, diagnostic: EvalRunFailureDiagnostic | None=None) -> EvalRunRecord

Terminalize owned work with a closed, credential-free diagnostic.

finish_cancel (method)

async EvalStore.finish_cancel(claim: EvalRunClaim) -> EvalRunRecord

Acknowledge cancellation after the owner has stopped execution.

release_run (method)

async EvalStore.release_run(claim: EvalRunClaim) -> EvalRunRecord

Release stopped owned work for retry, or finish requested cancellation.

load_result (method)

async EvalStore.load_result(run_id: str, *, max_bytes: int=CORPUS_EXECUTION_RESULT_MAX_BYTES) -> CorpusExecutionResult | None

Load one immutable safe result under an exact byte ceiling.

save_captured_result (method)

async EvalStore.save_captured_result(corpus: EvalCorpusDocument, result: CapturedEvaluationResultV1, *, redact_json: Callable[[Any], Any]) -> EvalResultRecord

Atomically save one immutable corpus and its captured result.

Existing custom V1 stores remain constructible. They opt into the new contract by overriding this method and captured_results.

load_result_by_revision (method)

async EvalStore.load_result_by_revision(revision: str, *, max_bytes: int=CORPUS_EXECUTION_RESULT_MAX_BYTES) -> CorpusExecutionResult | CapturedEvaluationResultV1 | None

Load one immutable result independent of its captured/fresh origin.

load_result_record (method)

async EvalStore.load_result_record(revision: str) -> EvalResultRecord | None

Load bounded metadata for one immutable result revision.

list_results (method)

async EvalStore.list_results(query: EvalResultQuery) -> EvalResultPage

List target-scoped immutable result metadata in keyset order.

set_baseline (method)

async EvalStore.set_baseline(update: EvalBaselineUpdate, *, redact_json: Callable[[Any], Any]) -> EvalBaselineMutationRecord

Commit or replay one actor-attributed baseline CAS mutation.

load_baseline (method)

async EvalStore.load_baseline(key: EvalBaselineKey) -> EvalBaselineRecord | None

Load the current baseline pointer for one exact evaluation scope.

load_baseline_mutation (method)

async EvalStore.load_baseline_mutation(operation_id: str) -> EvalBaselineMutationRecord | None

Load one immutable baseline audit fact by its idempotency digest.

Release source · Runtime guarantees · Capability guides

`EvalStorePublicationRejected`

from cayu.evals import EvalStorePublicationRejected

Public eval data could not cross the active credential-redaction boundary.

Bases: ValueError.

This type’s implementation and inherited behavior are available in the linked release source.

Release source · Runtime guarantees · Capability guides

`EvalStoreResultTooLarge`

from cayu.evals import EvalStoreResultTooLarge

A bounded eval-store read cannot fit the requested byte ceiling.

Bases: ValueError.

init (method)

EvalStoreResultTooLarge.__init__(max_bytes: int) -> None

Release source · Runtime guarantees · Capability guides

`EvalStoreTransientContention`

from cayu.evals import EvalStoreTransientContention

A bounded durable eval-store write exhausted its transient contention budget.

Bases: RuntimeError.

This type’s implementation and inherited behavior are available in the linked release source.

Release source · Runtime guarantees · Capability guides

`EvalSuiteCatalogEntry`

from cayu.evals import EvalSuiteCatalogEntry

Bases: _EvalStoreModel.

Fields and attributes

Name	Type	Default
`corpus_revision`	`StrictStr`	`required`
`id`	`StrictStr`	`required`
`revision`	`StrictStr`	`required`
`name`	`StrictStr`	`required`
`description`	`StrictStr \| None`	`None`
`case_count`	`StrictInt`	`Field(ge=1, le=EVAL_CORPUS_MAX_CASES)`
`assertion_count`	`StrictInt`	`Field(ge=1, le=EVAL_CORPUS_MAX_CASES * EVAL_CORPUS_MAX_ASSERTIONS_PER_CASE)`
`trials`	`StrictInt`	`Field(ge=1, le=EVAL_CORPUS_MAX_TRIALS)`
`timeout_seconds`	`StrictInt`	`Field(ge=1, le=EVAL_CORPUS_MAX_TIMEOUT_SECONDS)`

validate_revision_fields (method)

EvalSuiteCatalogEntry.validate_revision_fields(value: str, info) -> str

validate_id (method)

EvalSuiteCatalogEntry.validate_id(value: str, info) -> str

validate_name (method)

EvalSuiteCatalogEntry.validate_name(value: str, info) -> str

validate_description (method)

EvalSuiteCatalogEntry.validate_description(value: str | None, info) -> str | None

validate_counts (method)

EvalSuiteCatalogEntry.validate_counts() -> EvalSuiteCatalogEntry

Release source · Runtime guarantees · Capability guides

`EvalSuiteCatalogPage`

from cayu.evals import EvalSuiteCatalogPage

Bases: _EvalStoreModel.

Fields and attributes

Name	Type	Default
`items`	`tuple[EvalSuiteCatalogEntry, ...]`	`Field(max_length=EVAL_STORE_MAX_PAGE_SIZE)`
`next_cursor`	`StrictStr \| None`	`None`
`has_more`	`StrictBool`	`False`

validate_page (method)

EvalSuiteCatalogPage.validate_page() -> EvalSuiteCatalogPage

Release source · Runtime guarantees · Capability guides

`EvalSuiteCatalogQuery`

from cayu.evals import EvalSuiteCatalogQuery

Bases: _EvalStoreModel.

Fields and attributes

Name	Type	Default
`corpus_revision`	`StrictStr`	`required`
`cursor`	`StrictStr \| None`	`None`
`limit`	`StrictInt`	`Field(default=EVAL_STORE_DEFAULT_PAGE_SIZE, ge=1, le=EVAL_STORE_MAX_PAGE_SIZE)`
`max_result_bytes`	`StrictInt`	`Field(default=EVAL_STORE_DEFAULT_PAGE_BYTES, ge=1024, le=EVAL_STORE_MAX_PAGE_BYTES)`

validate_revision (method)

EvalSuiteCatalogQuery.validate_revision(value: str, info) -> str

validate_cursor (method)

EvalSuiteCatalogQuery.validate_cursor(value: str | None, info) -> str | None

Release source · Runtime guarantees · Capability guides

`InMemoryEvalStore`

from cayu.evals import InMemoryEvalStore

Explicitly ephemeral EvalStore for tests and process-local SDK use.

Bases: EvalStore.

close (method)

async InMemoryEvalStore.close() -> None

save_corpus (method)

async InMemoryEvalStore.save_corpus(corpus: EvalCorpusDocument, *, redact_json: Callable[[Any], Any]) -> EvalCorpusCatalogEntry

load_corpus (method)

async InMemoryEvalStore.load_corpus(revision: str, *, max_bytes: int=EVAL_CORPUS_MAX_BYTES) -> EvalCorpusDocument | None

list_corpora (method)

async InMemoryEvalStore.list_corpora(query: EvalCatalogQuery | None=None) -> EvalCorpusCatalogPage

save_scenario (method)

async InMemoryEvalStore.save_scenario(scenario: EvalScenarioDocumentV2, *, redact_json: Callable[[Any], Any]) -> EvalScenarioCatalogEntry

load_scenario (method)

async InMemoryEvalStore.load_scenario(revision: str, *, max_bytes: int=EVAL_SCENARIO_MAX_BYTES) -> EvalScenarioDocumentV2 | None

list_scenarios (method)

async InMemoryEvalStore.list_scenarios(query: EvalScenarioCatalogQuery | None=None) -> EvalScenarioCatalogPage

save_authored_suite (method)

async InMemoryEvalStore.save_authored_suite(document: EvalSuiteDocument, *, redact_json: Callable[[Any], Any]) -> EvalAuthoredSuiteCatalogEntry

load_authored_suite (method)

async InMemoryEvalStore.load_authored_suite(revision: str, *, max_bytes: int=EVAL_SUITE_AUTHORING_MAX_BYTES) -> EvalSuiteDocument | None

list_authored_suites (method)

async InMemoryEvalStore.list_authored_suites(query: EvalAuthoredSuiteCatalogQuery | None=None) -> EvalAuthoredSuiteCatalogPage

save_judge_calibration (method)

async InMemoryEvalStore.save_judge_calibration(report: EvalJudgeCalibrationReportV1, *, redact_json: Callable[[Any], Any]) -> EvalJudgeCalibrationReportV1

load_judge_calibration (method)

async InMemoryEvalStore.load_judge_calibration(revision: str, *, max_bytes: int=EVAL_JUDGE_CALIBRATION_MAX_BYTES) -> EvalJudgeCalibrationReportV1 | None

load_judge_calibration_by_run_id (method)

async InMemoryEvalStore.load_judge_calibration_by_run_id(run_id: str, *, max_bytes: int=EVAL_JUDGE_CALIBRATION_MAX_BYTES) -> EvalJudgeCalibrationReportV1 | None

list_suites (method)

async InMemoryEvalStore.list_suites(query: EvalSuiteCatalogQuery) -> EvalSuiteCatalogPage

list_cases (method)

async InMemoryEvalStore.list_cases(query: EvalCaseCatalogQuery) -> EvalCaseCatalogPage

admit_run (method)

async InMemoryEvalStore.admit_run(request: EvalRunRequest, *, redact_json: Callable[[Any], Any]) -> EvalRunRecord

load_run_by_idempotency_key (method)

async InMemoryEvalStore.load_run_by_idempotency_key(idempotency_key: str) -> EvalRunRecord | None

load_run (method)

async InMemoryEvalStore.load_run(run_id: str) -> EvalRunRecord | None

load_run_observation (method)

async InMemoryEvalStore.load_run_observation(run_id: str) -> EvalRunObservation | None

wait_for_run_terminal (method)

async InMemoryEvalStore.wait_for_run_terminal(run_id: str, *, timeout_seconds: float=30.0, poll_interval_seconds: float=EVAL_RUN_MIN_OBSERVATION_INTERVAL_SECONDS, max_poll_interval_seconds: float=EVAL_RUN_MAX_OBSERVATION_INTERVAL_SECONDS) -> EvalRunObservation | None

Wait within a strict deadline for a terminal run observation.

Polling begins no faster than the public 50 ms floor and exponentially backs off to the caller’s bounded ceiling. A missing run returns None; deadline exhaustion raises TimeoutError.

list_runs (method)

async InMemoryEvalStore.list_runs(query: EvalRunQuery | None=None) -> EvalRunPage

claim_run (method)

async InMemoryEvalStore.claim_run(*, target_key: str | None=None, launch_revision: str | None=None, lease_seconds: int=300) -> EvalRunLease | None

claim_run_for_targets (method)

async InMemoryEvalStore.claim_run_for_targets(target_keys: tuple[str, ...], *, lease_seconds: int=300) -> EvalRunLease | None

heartbeat_run (method)

async InMemoryEvalStore.heartbeat_run(claim: EvalRunClaim, *, extend_seconds: int=300) -> EvalRunRecord

heartbeat_run_observation (method)

async InMemoryEvalStore.heartbeat_run_observation(claim: EvalRunClaim, *, extend_seconds: int=300) -> EvalRunObservation

request_cancel (method)

async InMemoryEvalStore.request_cancel(run_id: str) -> EvalRunRecord

initialize_scenario_progress (method)

async InMemoryEvalStore.initialize_scenario_progress(claim: EvalRunClaim, progress: EvalScenarioRunProgress) -> EvalRunRecord

update_scenario_trial (method)

async InMemoryEvalStore.update_scenario_trial(claim: EvalRunClaim, trial: EvalScenarioTrialProgress) -> EvalRunRecord

load_trial_checkpoints (method)

async InMemoryEvalStore.load_trial_checkpoints(claim: EvalRunClaim) -> tuple[EvalRunTrialCheckpoint, ...]

load_trial_evidence_links (method)

async InMemoryEvalStore.load_trial_evidence_links(run_id: str) -> tuple[EvalTrialEvidenceLinkV1, ...]

save_trial_checkpoint (method)

async InMemoryEvalStore.save_trial_checkpoint(claim: EvalRunClaim, checkpoint: EvalRunTrialCheckpoint, *, redact_json: Callable[[Any], Any]) -> None

submit_scenario_approval (method)

async InMemoryEvalStore.submit_scenario_approval(run_id: str, submission: EvalScenarioApprovalSubmission) -> EvalRunRecord

publish_result (method)

async InMemoryEvalStore.publish_result(claim: EvalRunClaim, result: CorpusExecutionResult, *, redact_json: Callable[[Any], Any]) -> EvalRunRecord

fail_run (method)

async InMemoryEvalStore.fail_run(claim: EvalRunClaim, code: EvalRunFailureCode, *, diagnostic: EvalRunFailureDiagnostic | None=None) -> EvalRunRecord

finish_cancel (method)

async InMemoryEvalStore.finish_cancel(claim: EvalRunClaim) -> EvalRunRecord

release_run (method)

async InMemoryEvalStore.release_run(claim: EvalRunClaim) -> EvalRunRecord

load_result (method)

async InMemoryEvalStore.load_result(run_id: str, *, max_bytes: int=CORPUS_EXECUTION_RESULT_MAX_BYTES) -> CorpusExecutionResult | None

save_captured_result (method)

async InMemoryEvalStore.save_captured_result(corpus: EvalCorpusDocument, result: CapturedEvaluationResultV1, *, redact_json: Callable[[Any], Any]) -> EvalResultRecord

load_result_by_revision (method)

async InMemoryEvalStore.load_result_by_revision(revision: str, *, max_bytes: int=CORPUS_EXECUTION_RESULT_MAX_BYTES) -> CorpusExecutionResult | CapturedEvaluationResultV1 | None

load_result_record (method)

async InMemoryEvalStore.load_result_record(revision: str) -> EvalResultRecord | None

list_results (method)

async InMemoryEvalStore.list_results(query: EvalResultQuery) -> EvalResultPage

set_baseline (method)

async InMemoryEvalStore.set_baseline(update: EvalBaselineUpdate, *, redact_json: Callable[[Any], Any]) -> EvalBaselineMutationRecord

load_baseline (method)

async InMemoryEvalStore.load_baseline(key: EvalBaselineKey) -> EvalBaselineRecord | None

load_baseline_mutation (method)

async InMemoryEvalStore.load_baseline_mutation(operation_id: str) -> EvalBaselineMutationRecord | None

init (method)

InMemoryEvalStore.__init__(*, clock: Callable[[], datetime] | None=None, claim_id_factory: Callable[[], str] | None=None) -> None

Release source · Runtime guarantees · Capability guides

`TERMINAL_EVAL_RUN_STATUSES`

from cayu.evals import TERMINAL_EVAL_RUN_STATUSES

TERMINAL_EVAL_RUN_STATUSES = frozenset({EvalRunStatus.COMPLETED, EvalRunStatus.FAILED, EvalRunStatus.CANCELLED})

Release source · Runtime guarantees · Capability guides