Module: scenarios - Bijection

Classes

References

Run

Re-exports Run

RunCapture

Re-exports RunCapture

RunClient

Re-exports RunClient

Type Aliases

ScenarioRetention

Ƭ ScenarioRetention: Object Independent real-time lifetimes from admission, bounded by inherited state.

Type declaration

Name	Type	Description
`outputHours`	`number`	Zero retains no output; otherwise 1–72 hours.
`evidenceHours`	`number`	Reconstruct authorized report values for 1–720 hours.

ScenarioPlan

Ƭ ScenarioPlan: Object

Type declaration

Name	Type	Description
`simulation?`	`ScenarioSimulation`	-
`from?`	`SnapshotRef`	-
`start`	`ScenarioStart`	-
`programId`	`string` \| `null`	-
`recipe`	`ScenarioRecipe`	-
`histories`	`Record`<`string`, `ScenarioHistory`>	-
`program`	`ProgramReference`	-
`replacements`	`Record`<`string`, `string`>	Exact model replacements applied to candidate decisions only; fixed environment models and evaluators keep the original program’s models.

ScenarioSimulation

Ƭ ScenarioSimulation: Object

Type declaration

Name	Type
`startMs`	`number`
`seed`	`string`

ScenarioRecipe

Ƭ ScenarioRecipe: Object

Type declaration

Name	Type	Description
`program?`	`ScenarioProgramDefinition`	-
`modelBudget?`	`ScenarioModelBudget`	Work each inference step’s model tree may perform. Omit for the default allowance.
`observations?`	`ScenarioObservationScopes`	-
`name`	`string`	-
`caseName`	`string`	-
`inputs`	`unknown`	-
`inputsValidator`	`unknown`	-
`steps`	`ScenarioStep`[]	-

ScenarioProgramDefinition

Ƭ ScenarioProgramDefinition: Object Not exported: an exported type refers to it.

Type declaration

Name	Type
`function`	`ScenarioFunction`
`maxSteps`	`number`
`measurement?`	`ScenarioProgramMeasurement`

ScenarioProgramMeasurement

Ƭ ScenarioProgramMeasurement: Object Not exported: an exported type refers to it.

Type declaration

Name	Type
`id`	`string`
`function`	`ScenarioFunction`

ScenarioModelBudget

Ƭ ScenarioModelBudget: Object The complete work of one inference step’s model tree: every nested program call and fan-out child is charged to the same evaluation. Per-value byte bounds and the run record bound still apply.

Type declaration

Name	Type	Description
`calls`	`number`	Model invocations, local and provider (1–1024).
`providerCalls`	`number`	Provider dispatches, at most `calls` (1–64).

ScenarioObservationScopes

Ƭ ScenarioObservationScopes: Object Independent read ceilings for experimental roles; existing row, field, marking and live-authority checks still apply.

Type declaration

Name	Type
`candidate?`	`ScenarioObservationCeiling`[]
`environment?`	`ScenarioObservationCeiling`[]
`evaluator?`	`ScenarioObservationCeiling`[]

ScenarioObservationCeiling

Ƭ ScenarioObservationCeiling: Object Not exported: an exported type refers to it.

Type declaration

Name	Type
`relation`	`string`
`fields?`	`Record`<`string`, `string`>
`when?`	`unknown`

ScenarioStep

Ƭ ScenarioStep: { kind: "operation" ; id: string ; path: string ; invoke: ScenarioFunction ; recover: ScenarioFunction ; contractDigest: string ; args: ScenarioExpression } | { kind: "observation" ; id: string ; source: ScenarioSource ; history: string ; decisionAt: number ; cutoff: number } | { kind: "assumption" ; id: string ; operation: string ; states?: ScenarioAssumedStates } | { kind: "query" ; id: string ; function: ScenarioFunction ; args: ScenarioExpression } | { kind: "measurement" ; id: string ; function: ScenarioFunction ; args: ScenarioExpression } | { kind: "expectation" ; id: string ; actual: ScenarioExpression ; comparison: ScenarioComparison ; expected: ScenarioExpression } | { kind: "refusal" ; id: string ; operation: string ; code: string } | { kind: "snapshot" ; id: string ; ttlHours: number } | { kind: "advance" ; id: string ; byMs: number ; maxJobs: number } | { kind: "inference" ; id: string ; model: string ; binding: ScenarioModelBinding ; args: ScenarioExpression } Not exported: an exported type refers to it.

ScenarioSource

Ƭ ScenarioSource: Object

Type declaration

Name	Type
`module`	`string`
`export`	`string`
`collection`	`string`

ScenarioAssumedStates

Ƭ ScenarioAssumedStates: Object A literal hypothetical alternative, checked against the named source and the held command’s declared field scope. It carries no delivery evidence.

Type declaration

Name	Type
`source`	`ScenarioSource`
`values`	`Record`<`string`, `unknown`>

ScenarioComparison

Ƭ ScenarioComparison: "equal" | "at_least" | "at_most" Not exported: an exported type refers to it.

ScenarioModelBinding

Ƭ ScenarioModelBinding: "candidate" | "fixed"

ScenarioHistory

Ƭ ScenarioHistory: Object

Type declaration

Name	Type	Description
`observations`	`ScenarioObservation`[]	-
`producer?`	`ScenarioHistoryProducer`	The model version that computed these observations outside the run, such as a forecast over earlier history. Absent for supplied facts.

ScenarioObservation

Ƭ ScenarioObservation: Object

Type declaration

Name	Type	Description
`id`	`string`	-
`key`	`string`	-
`knownAt`	`number`	-
`validFrom`	`number`	-
`record`	`unknown`	A complete source record, or null for explicit absence.

ScenarioHistoryProducer

Ƭ ScenarioHistoryProducer: Object Not exported: an exported type refers to it.

Type declaration

Name	Type
`model`	`string`
`version`	`string`

ScenarioProgramCaptureRequest

Ƭ ScenarioProgramCaptureRequest: Object

Type declaration

Name	Type	Description
`programId`	`string`	-
`source?`	`BranchReference`	Omit to capture the live application without creating a source branch.
`cases`	`ScenarioProgramCase`[]	-

ScenarioProgramRequest

Ƭ ScenarioProgramRequest: Object

Type declaration

Name	Type
`programId`	`string`

ScenarioProgramStatus

Ƭ ScenarioProgramStatus: Object

Type declaration

Name	Type
`programId`	`string`
`cases`	`ScenarioProgramCase`[]
`program`	`ProgramReference`
`progress`	`ScenarioProgramProgress`
`deadline`	`string`

ScenarioProgramProgress

Ƭ ScenarioProgramProgress: { state: "capturing" } | { state: "ready" } | { state: "expired" } | { state: "failed" ; code: string }

ScenarioProgramRetirementResult

Ƭ ScenarioProgramRetirementResult: Object

Type declaration

Name	Type
`status`	`"retired"`
`programId`	`string`

ScenarioPlanRequest

Ƭ ScenarioPlanRequest: Object Not exported: an exported type refers to it.

Type declaration

Name	Type
`simulation?`	`ScenarioSimulation` \| `null`
`from?`	`SnapshotRef` \| `null`
`start?`	`ScenarioStart`
`release?`	`ReleaseRef` \| `null`
`name`	`string`
`caseName`	`string`
`histories?`	`Record`<`string`, `ScenarioHistory`>

ModelRunPlanRequest

Ƭ ModelRunPlanRequest: Object Planning resolves a handle in the selected executable at the selected state.

Type declaration

Name	Type
`model`	`ModelReference`
`input`	`unknown`
`from?`	`SnapshotRef` \| `null`
`release?`	`ReleaseRef` \| `null`
`simulation`	`ScenarioSimulation`
`reuse?`	`string` \| `null`
`budget?`	`ScenarioModelBudget` \| `null`

ModelReference

Ƭ ModelReference: Object Not exported: an exported type refers to it.

Type declaration

Name	Type
`format`	`string`
`name`	`string`
`input`	`unknown`
`output`	`unknown`

ScenarioRunRequest

Ƭ ScenarioRunRequest: Object Not exported: an exported type refers to it.

Type declaration

Name	Type
`runId`	`string`

ScenarioStatus

Ƭ ScenarioStatus: Object

Type declaration

Name	Type
`output?`	`SnapshotRef`
`runId`	`string`
`branch`	`BranchReference` \| `null`
`branchState`	`BranchStateName` \| `null`
`execution`	`ScenarioExecutionState`
`settlement`	`ScenarioSettlement`
`cleanup`	`ScenarioCleanup`
`deadline`	`string`
`outputExpiresAt?`	`string`
`evidenceExpiresAt?`	`string`

ScenarioExecutionState

Ƭ ScenarioExecutionState: { state: "admitted" } | { state: "running" } | { state: "waiting" } | { state: "completed" } | { state: "failed" } | { state: "cancelled" } Control state excludes step labels, refusal codes and business outcomes.

ScenarioSettlement

Ƭ ScenarioSettlement: { state: "pending" } | { state: "settled" } | { state: "unknown" } Whether a stopped run has reconciled every possibly submitted operation.

ScenarioCleanup

Ƭ ScenarioCleanup: { state: "pending" } | { state: "scheduled" ; reclaimAt: string } | { state: "reclaiming" } | { state: "complete" } | { state: "retained" ; expiresAt: string } | { state: "blocked" ; message: string }

ScenarioReport

Ƭ ScenarioReport: Object

Type declaration

Name	Type
`result`	`ScenarioResult`
`snapshots?`	`Record`<`string`, `SnapshotRef`>
`output?`	`SnapshotRef`
`runId`	`string`
`name`	`string`
`caseName`	`string`
`branch`	`BranchReference` \| `null`
`branchState`	`BranchStateName` \| `null`
`program`	`ProgramReference`
`startingStateRevision`	`string`
`inputs`	`ScenarioRunInputs`
`execution`	`ScenarioExecutionState`
`settlement`	`ScenarioSettlement`
`cleanup`	`ScenarioCleanup`
`checks`	`ScenarioCheckReport`[]
`assumptions`	`ScenarioAssumptionReport`[]
`observations`	`ScenarioObservationReport`[]
`measurements`	`ScenarioMeasurementReport`[]
`inferences`	`ScenarioInferenceReport`[]
`completedSteps`	`number`
`passedChecks`	`number`
`totalChecks`	`number`
`totalSteps`	`number`
`deadline`	`string`
`outputExpiresAt?`	`string`
`evidenceExpiresAt?`	`string`

ScenarioResult

Ƭ ScenarioResult: { state: "pending" } | { state: "value" ; value: unknown } | { state: "unavailable" } A run result is disclosed only after its dependencies are established under the reader’s current authority. JSON preserves exact Bijection scalars.

ScenarioRunInputs

Ƭ ScenarioRunInputs: Object What a run was started from and with, without business values: its starting snapshot and behavior, sampling controls, candidate program, replaced decisions, captured history names and sibling.

Type declaration

Name	Type	Description
`start`	`ScenarioStart`	-
`from?`	`SnapshotRef`	-
`simulation?`	`ScenarioSimulation`	-
`programId`	`string` \| `null`	-
`replacements`	`string`[]	`model.dependency...` paths whose candidate models the plan replaces.
`histories`	`string`[]	Names of the captured observation histories.
`siblingOf`	`string` \| `null`	The run whose starting state this run shares.
`collectionId`	`string` \| `null`	The experiment that admitted this run.
`specificationDigest`	`string`	SHA-256 of the plan with its sampling seed removed. Two runs with the same digest and seed recompute one specification; with different seeds they resample it.

ScenarioCheckReport

Ƭ ScenarioCheckReport: Object

Type declaration

Name	Type
`id`	`string`
`outcome`	`ScenarioCheckOutcome`

ScenarioCheckOutcome

Ƭ ScenarioCheckOutcome: { state: "pending" } | { state: "passed" ; actual: unknown ; expected: unknown } | { state: "failed" ; actual: unknown ; expected: unknown } | { state: "withheld" } | { state: "unavailable" }

ScenarioAssumptionReport

Ƭ ScenarioAssumptionReport: Object

Type declaration

Name	Type
`id`	`string`
`operation`	`string`
`consequence`	`ScenarioAssumptionConsequence`
`outcome`	`ScenarioAssumptionOutcome`

ScenarioAssumptionConsequence

Ƭ ScenarioAssumptionConsequence: "declared_states" | "modeled_states" Not exported: an exported type refers to it.

ScenarioAssumptionOutcome

Ƭ ScenarioAssumptionOutcome: { state: "pending" } | { state: "admitted" ; calls: number ; digest: string } | { state: "withheld" } | { state: "unavailable" }

ScenarioObservationReport

Ƭ ScenarioObservationReport: Object

Type declaration

Name	Type
`id`	`string`
`history`	`string`
`provenance`	`ScenarioObservationProvenance`
`producer`	`ScenarioHistoryProducer` \| `null`
`outcome`	`ScenarioObservationOutcome`

ScenarioObservationProvenance

Ƭ ScenarioObservationProvenance: "simulated" | "predicted" Not exported: an exported type refers to it.

ScenarioObservationOutcome

Ƭ ScenarioObservationOutcome: { state: "pending" } | { state: "admitted" ; records: number ; digest: string } | { state: "withheld" } | { state: "unavailable" }

ScenarioMeasurementReport

Ƭ ScenarioMeasurementReport: Object

Type declaration

Name	Type
`id`	`string`
`outcome`	`ScenarioMeasurementOutcome`

ScenarioMeasurementOutcome

Ƭ ScenarioMeasurementOutcome: { state: "pending" } | { state: "measured" ; program: ProgramReference ; revision: string ; readings: unknown ; digest: string } | { state: "withheld" } | { state: "unavailable" }

ScenarioInferenceReport

Ƭ ScenarioInferenceReport: Object

Type declaration

Name	Type
`id`	`string`
`model`	`string`
`outcome`	`ScenarioInferenceOutcome`

ScenarioInferenceOutcome

Ƭ ScenarioInferenceOutcome: { state: "pending" } | { state: "inferred" ; version: string ; digest: string ; host: ProgramReference ; calls: ScenarioInferenceCall[] ; basis: ScenarioInferenceBasis } | { state: "unavailable" }

ScenarioInferenceCall

Ƭ ScenarioInferenceCall: Object

Type declaration

Name	Type
`path`	`string`
`model`	`string`
`version`	`string`
`inputDigest`	`string`
`outputDigest`	`string`

ScenarioInferenceBasis

Ƭ ScenarioInferenceBasis: Object Not exported: an exported type refers to it.

Type declaration

Name	Type
`program`	`ProgramReference`
`revision`	`string`
`observations?`	`ScenarioObservationCeiling`[]

ScenarioResultResponse

Ƭ ScenarioResultResponse: Object

Type declaration

Name	Type
`run_id`	`string`
`execution`	`ScenarioExecutionState`
`result`	`ScenarioResult`

ScenarioCollectionStartRequest

Ƭ ScenarioCollectionStartRequest: Object

Type declaration

Name	Type
`retention?`	`ScenarioRetention`
`budget?`	`ScenarioCollectionBudget`
`collectionId`	`string`
`cases`	`ScenarioCollectionCase`[]
`retain`	`boolean`

ScenarioCollectionBudget

Ƭ ScenarioCollectionBudget: Object Aggregate limits fixed before any collection child starts. The parent accounts every child’s admitted work; the native scheduler queues children while the collection is at its branch, byte or concurrency allowance.

Type declaration

Name	Type	Description
`concurrency`	`number`	Children executing at once (1–16).
`branches`	`number`	Branches the collection holds at once, its anchor and retained outputs included (1 to the owner’s open-branch limit).
`maxRuns`	`number`	Child runs (1–64, the owner’s retained run custody).
`maxWork`	`number`	Admitted execution work across children: each child’s step budget, scheduled jobs and model invocations.
`maxRetainedBytes`	`number`	Branch storage reservation held at once, 8 GiB per held branch.
`maxDurationMs`	`number`	Real-time deadline from admission (1–1800000 milliseconds).

ScenarioCollectionCase

Ƭ ScenarioCollectionCase: Object Not exported: an exported type refers to it.

Type declaration

Name	Type
`runId`	`string`
`plan`	`ScenarioPlan`

ScenarioCollectionRequest

Ƭ ScenarioCollectionRequest: Object Not exported: an exported type refers to it.

Type declaration

Name	Type
`collectionId`	`string`

ScenarioCollectionStatus

Ƭ ScenarioCollectionStatus: Object

Type declaration

Name	Type
`collectionId`	`string`
`execution`	`ScenarioCollectionExecution`
`admittedCases`	`number`
`totalCases`	`number`
`runs`	`ScenarioStatus`[]
`selection`	`ScenarioSelection` \| `null`
`training`	`ScenarioSelectionReference` \| `null`
`deadline`	`string`

ScenarioCollectionExecution

Ƭ ScenarioCollectionExecution: { state: "running" } | { state: "waiting" } | { state: "completed" } | { state: "cancelled" } Not exported: an exported type refers to it.

ScenarioSelection

Ƭ ScenarioSelection: Object Immutable identity of a native training decision; no metric values are cached.

Type declaration

Name	Type	Description
`collectionId`	`string`	-
`measurementId`	`string`	-
`runIds`	`string`[]	The selected candidate’s run for each aligned training case.
`programId`	`string`	-
`program`	`ProgramReference`	-
`replacements`	`Record`<`string`, `string`>	The selected candidate’s exact model replacements. A holdout runs the same set; promotion installs only code without replacements.
`inputsDigests`	`string`[]	Sorted distinct digests of the training cases’ inputs.
`historiesDigests`	`string`[]	Sorted distinct digests of the training cases’ captured histories.
`trainingDigest`	`string`	-
`evaluatorDigest`	`string`	-
`evaluatorProgram?`	`ProgramReference`	-
`startingProgram`	`ProgramReference`	-
`startingStateRevision`	`string`	-
`selectedAt`	`string`	-

ScenarioSelectionReference

Ƭ ScenarioSelectionReference: Object

Type declaration

Name	Type
`collectionId`	`string`
`trainingDigest`	`string`

ScenarioCollectionReport

Ƭ ScenarioCollectionReport: Object

Type declaration

Name	Type
`collectionId`	`string`
`execution`	`ScenarioCollectionExecution`
`admittedCases`	`number`
`totalCases`	`number`
`reports`	`ScenarioReport`[]
`selection`	`ScenarioSelection` \| `null`
`training`	`ScenarioSelectionReference` \| `null`
`deadline`	`string`

ScenarioCollectionOrderRequest

Ƭ ScenarioCollectionOrderRequest: Object Order a bounded collection’s candidates by one named frozen measurement.

Type declaration

Name	Type
`collectionId`	`string`
`measurementId`	`string`

ScenarioCollectionOrder

Ƭ ScenarioCollectionOrder: Object

Type declaration

Name	Type
`collectionId`	`string`
`measurementId`	`string`
`candidates`	`ScenarioCandidate`[]
`comparison`	`ScenarioCandidateComparison`

ScenarioCandidate

Ƭ ScenarioCandidate: Object Not exported: an exported type refers to it.

Type declaration

Name	Type
`candidate`	`string`
`programId`	`string` \| `null`
`program`	`ProgramReference`
`replacements`	`Record`<`string`, `string`>
`runIds`	`string`[]

ScenarioCandidateComparison

Ƭ ScenarioCandidateComparison: { outcome: "preferred" ; candidate: string ; ranking: ScenarioCandidateRanking } | { outcome: "tie" ; candidates: string[] ; ranking: ScenarioCandidateRanking } | { outcome: "no_feasible" ; ranking: ScenarioCandidateRanking } | { outcome: "incomplete" ; reason: string ; ranking: ScenarioCandidateRanking | null } | { outcome: "incomparable" ; reason: string } Not exported: an exported type refers to it.

ScenarioCandidateRanking

Ƭ ScenarioCandidateRanking: Object Not exported: an exported type refers to it.

Type declaration

Name	Type
`objective`	`unknown`
`cases`	`number`
`places`	`ScenarioCandidatePlace`[]
`notComparable`	`ScenarioCandidateNotComparable`[]

ScenarioCandidatePlace

Ƭ ScenarioCandidatePlace: Object Not exported: an exported type refers to it.

Type declaration

Name	Type
`place`	`number`
`candidates`	`string`[]
`withinLimits`	`boolean`
`missed`	`string`[]

ScenarioCandidateNotComparable

Ƭ ScenarioCandidateNotComparable: Object Not exported: an exported type refers to it.

Type declaration

Name	Type
`candidate`	`string`
`reasons`	`string`[]

ScenarioSelectionStatus

Ƭ ScenarioSelectionStatus: Object

Type declaration

Name	Type
`collectionId`	`string`
`selection`	`ScenarioSelection` \| `null`

ScenarioHoldoutStartRequest

Ƭ ScenarioHoldoutStartRequest: Object

Type declaration

Name	Type
`training`	`ScenarioSelectionReference`
`holdout`	`ScenarioCollectionStartRequest`

ScenarioPromotionRequest

Ƭ ScenarioPromotionRequest: Object

Type declaration

Name	Type	Description
`training`	`ScenarioSelectionReference`	-
`holdout`	`string`	The exact linked holdout whose current authorized evidence qualifies the selected program. A training winner alone cannot qualify code.
`key`	`string`	-

ScenarioPromotionProgress

Ƭ ScenarioPromotionProgress: { state: "pending" } | { state: "ready" } | { state: "expired" } | { state: "failed" ; code: string }

ScenarioPromotionStatus

Ƭ ScenarioPromotionStatus: Object

Type declaration

Name	Type
`collectionId`	`string`
`branch`	`BranchReference`
`program`	`ProgramReference`
`progress`	`ScenarioPromotionProgress`

SubmissionEnvelope

Ƭ SubmissionEnvelope: Object Not exported: an exported type refers to it.

Type declaration

Name	Type
`deployment`	`string`
`caller`	`OperationRequestScope`[`"caller"`] \| `null`
`endpoint`	`string`
`request`	`unknown`

ScenarioCall

Type declaration

▸ (endpoint, body, signal?): Promise<unknown>

Parameters

Name	Type
`endpoint`	`"programs/capture"` \| `"programs/status"` \| `"programs/retire"` \| `"programs/promote"` \| `"programs/promotion"` \| `"models/list"` \| `"models/plan"` \| `"models/inspect"` \| `"plan"` \| `"start"` \| `"status"` \| `"report"` \| `"result"` \| `"request"` \| `"resume"` \| `"cancel"` \| `"retire"` \| `"collections/start"` \| `"collections/status"` \| `"collections/report"` \| `"collections/order"` \| `"collections/select"` \| `"collections/selection"` \| `"collections/holdout"` \| `"collections/resume"` \| `"collections/cancel"` \| `"collections/retire"`
`body`	`ScenarioProgramCaptureRequest` \| `ScenarioProgramRequest` \| `ScenarioPromotionRequest` \| `Record`<`string`, `never`> \| `ModelRunPlanRequest` \| `ScenarioPlanRequest` \| `ScenarioStartRequest` \| `ScenarioRunRequest` \| `ScenarioCollectionStartRequest` \| `ScenarioCollectionRequest` \| `ScenarioCollectionOrderRequest` \| `ScenarioHoldoutStartRequest`
`signal?`	`AbortSignal`

Returns

Promise<unknown>

ScenarioCollectionCapture

Ƭ ScenarioCollectionCapture: Object Persist this complete value before admission, including the holdout link. Restore it to retry an uncertain start without replanning or minting IDs.

Type declaration

Name	Type
`request`	`Readonly`<`ScenarioCollectionStartRequest`>
`training`	`Readonly`<`ScenarioSelectionReference`> \| `null`

ScenarioInput

Ƭ ScenarioInput<T>: ScenarioValue<T> | T extends readonly infer Item[] ? ScenarioInput<Item>[] : T extends ArrayBuffer ? T : T extends object ? { [Key in keyof T]: ScenarioInput<T[Key]> } : T Arguments may refer to earlier results without losing their declared types.

Type parameters

Name
`T`

ScenarioOperation

Ƭ ScenarioOperation<Result>: Object

Type parameters

Name
`Result`

Type declaration

Name	Type
`id`	`string`
`result`	`ScenarioValue`<`Result`>
`outcome`	`ScenarioOutcome`

ScenarioObservationEntry

Ƭ ScenarioObservationEntry: string | { relation: string ; fields?: Record<string, AccessClaimType> ; when?: AccessCondition } One relation a role may read: its name, or the relation with the only fields it may read (typed for when) and a condition each read’s index range must prove. when reads row fields as resource.<field>, the simulated decision time as context.now and scalar case arguments as context.<argument>. Reads must .select(...) declared fields; a filter never establishes observation.

ScenarioObservationDeclaration

Ƭ ScenarioObservationDeclaration: Object

Type declaration

Name	Type
`candidate?`	readonly `ScenarioObservationEntry`[]
`environment?`	readonly `ScenarioObservationEntry`[]
`evaluator?`	readonly `ScenarioObservationEntry`[]

ScenarioDefinition

Ƭ ScenarioDefinition: Object

Type declaration

Name	Type
`kind`	`"scenario"`
`caseNames`	readonly `string`[]
`build`	(`name`: `string`, `caseName`: `string`) => `Promise`<`ScenarioRecipe`>

RunRelation

Ƭ RunRelation: { kind: "same" } | { kind: "captured" ; from: SnapshotRef ; step: string } | { kind: "fresh" ; from: SnapshotRef ; step: string } | { kind: "recompute" } | { kind: "resample" } | { kind: "same-specification" } | { kind: "sibling" } | { kind: "unrelated" } How a later run relates to an earlier one, from their native reports.

ScenarioProgramContext

Ƭ ScenarioProgramContext: Object

Type declaration

Name	Type
`scope`	(…`path`: (`string` \| `number`)[]) => `ScenarioProgramContext`
`call`	<I, O>(`key`: `string`, `procedure`: `ScenarioProcedure`<`I`, `O`>, `input`: `I`) => `Promise`<`O`>
`now`	() => `number`
`snapshot`	(`key`: `string`, `options?`: { `ttlHours?`: `number` }) => `Promise`<`SnapshotRef`>
`decide`	<I, O>(`key`: `string`, `model`: `ModelDefinition`<`I`, `O`>, `input`: `I`) => `Promise`<`O`>
`environment`	<I, O>(`key`: `string`, `model`: `ModelDefinition`<`I`, `O`>, `input`: `I`) => `Promise`<`O`>
`query`	<Q>(`key`: `string`, `query`: `Q`, `input`: `Q`[`"_args"`]) => `Promise`<`Q`[`"_returnType"`]>
`apply`	<O>(`key`: `string`, `operation`: `O`, `input`: `O`[`"_args"`]) => `Promise`<`O`[`"_returnType"`]>
`assume`	(`key`: `string`, `operation`: `string`) => `Promise`<`void`>
`assumeStates`	<S>(`key`: `string`, `operation`: `string`, `source`: `SourceHandle`<`S`>, `values`: `Partial`<`NonNullable`<`SourceHandle`<`S`>[`"__sourceType"`]>>) => `Promise`<`void`>
`advanceBy`	(`durationMs`: `number`, `options`: { `key`: `string` ; `maxJobs?`: `number` }) => `Promise`<`void`>
`measure`	<Q>(`key`: `string`, `query`: `Q`, `input`: `Q`[`"_args"`]) => `Promise`<`Q`[`"_returnType"`]>
`realize`	(`key`: `string`, `source`: `SourceHandle`, `prediction`: { `step`: `string` ; `path?`: `string`[] }, `options?`: { `decisionAt?`: `number` ; `cutoff?`: `number` }) => `Promise`<`void`>

ScenarioProcedure

Ƭ ScenarioProcedure<I, O>: Object A reusable sequence of scenario instructions with a typed input and result. Calling it from several places, or from another procedure, keeps every instruction key distinct through its scope.

Type parameters

Name
`I`
`O`

Type declaration

Name	Type
`kind`	`"scenarioProcedure"`

ScenarioSubmission

Ƭ ScenarioSubmission: Pick<SubmissionEnvelope, "deployment" | "caller"> & { endpoint: "start" ; request: ScenarioStartRequest } | { endpoint: "resume" | "cancel" | "retire" ; request: ScenarioRunRequest } | { endpoint: "collections/start" ; request: ScenarioCollectionStartRequest } | { endpoint: "collections/holdout" ; request: ScenarioHoldoutStartRequest } | { endpoint: "collections/resume" | "collections/cancel" | "collections/retire" ; request: ScenarioCollectionRequest } | { endpoint: "collections/select" ; request: ScenarioCollectionOrderRequest } | { endpoint: "programs/capture" ; request: ScenarioProgramCaptureRequest } | { endpoint: "programs/retire" ; request: ScenarioProgramRequest } | { endpoint: "programs/promote" ; request: ScenarioPromotionRequest }

ScenarioSubmissionScope

Ƭ ScenarioSubmissionScope: SubmissionScope<`scenario:${ScenarioSubmission[“endpoint”]}`>

ScenarioSubmissionStore

Ƭ ScenarioSubmissionStore: SubmissionStore<ScenarioSubmission, ScenarioSubmissionScope["ref"]>

AccessCondition

Ƭ AccessCondition<Resource, Principal, Context>: (scope: AccessConditionScope<Resource, Principal, Context>) => AccessExpression<boolean>

Type parameters

Name	Type
`Resource`	`any`
`Principal`	`any`
`Context`	`any`

Type declaration

▸ (scope): AccessExpression<boolean> Not exported: an exported type refers to it.

Parameters

Name	Type
`scope`	`AccessConditionScope`<`Resource`, `Principal`, `Context`>

Returns

AccessExpression<boolean>

Variables

MAX_SCENARIO_STEP_KEY_BYTES

• Const MAX_SCENARIO_STEP_KEY_BYTES: 256 Native step keys, including every enclosing program scope, admit this many bytes.

Functions

scenarioModelBudget

▸ scenarioModelBudget(budget): ScenarioModelBudget Check a model budget against the native bounds before admission.

Parameters

Name	Type
`budget`	`ScenarioModelBudget`

Returns

ScenarioModelBudget

defineScenario

▸ defineScenario<Args>(definition): ScenarioDefinition Define a versioned, repeatable recipe over ordinary business operations.

Type parameters

Name	Type
`Args`	extends `PropertyValidators`

Parameters

Name	Type	Description
`definition`	`Object`	-
`definition.args`	`Args`	-
`definition.observations?`	`ScenarioObservationDeclaration`	-
`definition.cases`	`Record`<`string`, `Infer`<`AsObjectValidator`<`Args`>>>	-
`definition.program?`	`FunctionReference`<`"query"`, `"public"`, `any`, `any`>	-
`definition.maxSteps?`	`number`	-
`definition.modelBudget?`	`ScenarioModelBudget`	Work each inference step’s complete model tree may perform, nested programs and `ctx.map` children included. Defaults to 64 calls, of which 16 provider calls.
`definition.measurement?`	`Object`	-
`definition.measurement.id`	`string`	-
`definition.measurement.query`	`FunctionReference`<`"query"`, `"public"`, `any`, `AppMetricReadings`>	-
`definition.build?`	(`builder`: `ScenarioBuilder`, `args`: `Infer`<`AsObjectValidator`<`Args`>>) => `void` \| `Promise`<`void`>	-

Returns

ScenarioDefinition

runRelation

▸ runRelation(earlier, later): RunRelation Classify how later relates to earlier. Checkpoints produced by the earlier run are its output (step: "output") and its named intermediate captures.

Parameters

Name	Type
`earlier`	`ScenarioReport`
`later`	`ScenarioReport`

Returns

RunRelation

defineScenarioProcedure

▸ defineScenarioProcedure<I, O>(run): ScenarioProcedure<I, O>

Type parameters

Name
`I`
`O`

Parameters

Name	Type
`run`	(`sim`: `ScenarioProgramContext`, `input`: `I`) => `Promise`<`O`>

Returns

ScenarioProcedure<I, O>

defineScenarioProgram

▸ defineScenarioProgram<A, O>(definition): RegisteredQuery<"public", { input: any ; answer: any ; references: any ; frame: null | string ; nowMs: number }, Promise<{ result?: undefined ; state: "call" ; step: string ; frame: string = encoded.frame; references: string[] = encoded.references } | { validator: string ; state: "complete" ; result: string ; step?: undefined ; frame?: undefined = encoded.frame; references?: undefined = encoded.references }>> Export the query and reference it from defineScenario({program}). Each call resumes the saved frame with the awaited instruction’s answer, runs one segment and returns the next instruction with its new frame.

Type parameters

Name	Type
`A`	extends `PropertyValidators`
`O`	extends `Value` = `null`

Parameters

Name	Type
`definition`	`Object`
`definition.args`	`A`
`definition.returns?`	`Validator`<`O`, `"required"`, `any`>
`definition.run`	(`sim`: `ScenarioProgramContext`, `input`: `Infer`<`AsObjectValidator`<`A`>>) => `Promise`<`O` \| `void`>

Returns

RegisteredQuery<"public", { input: any ; answer: any ; references: any ; frame: null | string ; nowMs: number }, Promise<{ result?: undefined ; state: "call" ; step: string ; frame: string = encoded.frame; references: string[] = encoded.references } | { validator: string ; state: "complete" ; result: string ; step?: undefined ; frame?: undefined = encoded.frame; references?: undefined = encoded.references }>>

createRunClient

▸ createRunClient(call, ownerSignal?): Readonly<{ start: <I, O>(spec: ModelRunStart<I, O>) => Promise<Run<O>><O>(spec: ScenarioRunStart) => Promise<Run<O>> ; get: <O>(run: string | RunCapture, options: { model?: ModelReference<any, O> }) => Run<O> ; list: (signal?: AbortSignal) => Promise<readonly ModelRunSummary[]> ; calibrate: <T>(definition: { name: string ; run: Run<T> }, signal?: AbortSignal) => Promise<ModelArtifact<T>> }> Start and observe scenario and model runs through one native lifecycle.

Parameters

Name	Type
`call`	`ScenarioCall`
`ownerSignal?`	`AbortSignal`

Returns

Readonly<{ start: <I, O>(spec: ModelRunStart<I, O>) => Promise<Run<O>><O>(spec: ScenarioRunStart) => Promise<Run<O>> ; get: <O>(run: string | RunCapture, options: { model?: ModelReference<any, O> }) => Run<O> ; list: (signal?: AbortSignal) => Promise<readonly ModelRunSummary[]> ; calibrate: <T>(definition: { name: string ; run: Run<T> }, signal?: AbortSignal) => Promise<ModelArtifact<T>> }>

​Classes

​References

​Run

​RunCapture

​RunClient

​ExperimentClient

​ScenarioRunStart

​ModelRunStart

​Type Aliases

​ScenarioRetention

​Type declaration

​ScenarioPlan

​Type declaration

​ScenarioSimulation

​Type declaration

​ScenarioRecipe

​Type declaration

​ScenarioProgramDefinition

​Type declaration

​ScenarioProgramMeasurement

​Type declaration

​ScenarioModelBudget

​Type declaration

​ScenarioObservationScopes

​Type declaration

​ScenarioObservationCeiling

​Type declaration

​ScenarioStep

​ScenarioSource

​Type declaration

​ScenarioAssumedStates

​Type declaration

​ScenarioComparison

​ScenarioModelBinding

​ScenarioHistory

​Type declaration

​ScenarioObservation

​Type declaration

​ScenarioHistoryProducer

​Type declaration

​ScenarioProgramCaptureRequest

​Type declaration

​ScenarioProgramRequest

​Type declaration

​ScenarioProgramStatus

​Type declaration

​ScenarioProgramProgress

​ScenarioProgramRetirementResult

​Type declaration

​ScenarioPlanRequest

​Type declaration

​ModelRunPlanRequest

​Type declaration

​ModelReference

​Type declaration

​ScenarioRunRequest

​Type declaration

​ScenarioStatus

​Type declaration

​ScenarioExecutionState

​ScenarioSettlement

​ScenarioCleanup

​ScenarioReport

​Type declaration

​ScenarioResult

​ScenarioRunInputs

​Type declaration

​ScenarioCheckReport

​Type declaration

​ScenarioCheckOutcome

​ScenarioAssumptionReport

​Type declaration

​ScenarioAssumptionConsequence

​ScenarioAssumptionOutcome

​ScenarioObservationReport

​Type declaration

​ScenarioObservationProvenance

​ScenarioObservationOutcome

​ScenarioMeasurementReport

​Type declaration

Classes

References

Run

RunCapture

RunClient

ExperimentClient

ScenarioRunStart

ModelRunStart

Type Aliases

ScenarioRetention

Type declaration

ScenarioPlan

Type declaration

ScenarioSimulation

Type declaration

ScenarioRecipe

Type declaration

ScenarioProgramDefinition

Type declaration

ScenarioProgramMeasurement

Type declaration

ScenarioModelBudget

Type declaration

ScenarioObservationScopes

Type declaration

ScenarioObservationCeiling

Type declaration

ScenarioStep

ScenarioSource

Type declaration

ScenarioAssumedStates

Type declaration

ScenarioComparison

ScenarioModelBinding

ScenarioHistory

Type declaration

ScenarioObservation

Type declaration

ScenarioHistoryProducer

Type declaration

ScenarioProgramCaptureRequest

Type declaration

ScenarioProgramRequest

Type declaration

ScenarioProgramStatus

Type declaration

ScenarioProgramProgress

ScenarioProgramRetirementResult

Type declaration

ScenarioPlanRequest

Type declaration

ModelRunPlanRequest

Type declaration

ModelReference

Type declaration

ScenarioRunRequest

Type declaration

ScenarioStatus

Type declaration

ScenarioExecutionState

ScenarioSettlement

ScenarioCleanup

ScenarioReport

Type declaration

ScenarioResult

ScenarioRunInputs

Type declaration

ScenarioCheckReport

Type declaration

ScenarioCheckOutcome

ScenarioAssumptionReport

Type declaration

ScenarioAssumptionConsequence

ScenarioAssumptionOutcome

ScenarioObservationReport

Type declaration

ScenarioObservationProvenance

ScenarioObservationOutcome

ScenarioMeasurementReport

Type declaration