ump_spec.models.decision

module ump-spec

Decision records and approvals (spec §12, §15, §18).

Classes

MaterialityInputs

class

class MaterialityInputs(StrictModel)

Fields

Name	Type	Default
`spend_at_stake`	`float`
`weeks_to_reverse`	`float`	`0`
`evoi`	`float \| None`	`None`
`client_recommendation`	`bool`	`False`
`public_claim`	`bool`	`False`

Source: packages/ump-spec/src/ump_spec/models/decision.py, line 18

Expected

class

class Expected(StrictModel)

Fields

Name	Type	Default
`median`	`float`
`interval`	`Interval`
`p_beats_current`	`float \| None`	`None`
`expected_regret`	`float \| None`	`None`

Source: packages/ump-spec/src/ump_spec/models/decision.py, line 26

Realised

class

class Realised(StrictModel)

Fields

Name	Type	Default
`value`	`float`
`date`	`date`
`within_curse_band`	`bool \| None`	`None`

Source: packages/ump-spec/src/ump_spec/models/decision.py, line 33

IntegrityRef

class

class IntegrityRef(StrictModel)

Fields

Name	Type	Default
`leaf_index`	`int \| None`	`None`
`leaf_hash`	`str \| None`	`None`
`domain_tag`	`str`	`'ump/decision/v1'`
`key_version`	`str \| None`	`None`
`checkpoint_size`	`int \| None`	`None`

Source: packages/ump-spec/src/ump_spec/models/decision.py, line 39

DecisionRecord

class

class DecisionRecord(ArtifactDoc)

Fields

Name	Type	Default	Description
`client_id`	`str \| None`	`None`
`question`	`str`
`options`	`list[str]`	`list()`	Scenario artifact ids.
`chosen_option`	`str \| None`	`None`
`comparison_estimand`	`Estimand \| None`	`None`
`expected`	`Expected \| None`	`None`
`headline`	`ShownNumber \| None`	`None`
`uplift_raw`	`float \| None`	`None`
`uplift_curse_adjusted`	`float \| None`	`None`
`curse_adjust_status`	`Literal['provisional', 'validated']`	`'provisional'`
`audit_level`	`Literal[1, 2, 3, 4]`	`2`
`materiality_tier`	`MaterialityTier \| None`	`None`
`materiality_inputs`	`MaterialityInputs \| None`	`None`
`gate_results`	`list[GateResult]`	`list()`
`fallbacks`	`list[FallbackEntry]`	`list()`
`drafted_by`	`str \| None`	`None`
`approved_by`	`str \| None`	`None`
`review_after`	`date \| None`	`None`
`realised`	`Realised \| None`	`None`
`integrity`	`IntegrityRef \| None`	`None`
`rationale`	`str \| None`	`None`
`model_id`	`str \| None`	`None`	Model the options were scored on. Written by planner (`service.py`) and Newton’s actor (`actor_tools.py`).
`model_version`	`int \| None`	`None`	Version of that model. Written by planner (`service.py`) and Newton’s actor (`actor_tools.py`).
`baseline_option`	`str \| None`	`None`	Scenario id of the baseline option. Written by planner (`service.py`) and Newton’s actor (`actor_tools.py`).
`baseline_plan`	`dict[str, Any] \| None`	`None`	Baseline spend per channel. Written by planner (`service.py`) and Newton’s actor (`actor_tools.py`).
`chosen_plan`	`dict[str, Any] \| None`	`None`	Chosen spend per channel. Written by planner (`service.py`) and Newton’s actor (`actor_tools.py`).
`crn_seed`	`int \| None`	`None`	Common-random-numbers seed shared by the compared options. Written by planner (`service.py`) and Newton’s actor (`actor_tools.py`).
`thinning_index`	`str \| None`	`None`	Which draws were used, e.g. `first:200`. Written by planner (`service.py`) and Newton’s actor (`actor_tools.py`).
`curse_adjustment`	`dict[str, Any] \| None`	`None`	Optimiser’s-curse shrinkage: status, method, shrink_factor (provisional until study 7). Written by planner (`service.py`) and Newton’s actor (`actor_tools.py`).
`dependence`	`dict[str, Any] \| str \| None`	`None`	Dependence declaration the comparison used (§8). Written by planner (`service.py`) and Newton’s actor (`actor_tools.py`).
`ladder_step`	`str \| None`	`None`	Fallback-ladder rung the numbers came from (§8). Written by planner (`service.py`) and Newton’s actor (`actor_tools.py`).
`materiality`	`dict[str, Any] \| None`	`None`	Materiality tier with reasons and the proposed thresholds (G-MAT-01). Written by planner (`service.py`) and Newton’s actor (`actor_tools.py`).
`review_after_default_days`	`dict[str, Any] \| None`	`None`	Default review window `{value, proposed}`. Written by planner (`service.py`) and Newton’s actor (`actor_tools.py`).
`seed_key`	`str \| None`	`None`	Seed-run key that made this record (e2e seed only). Written by planner (`service.py`) and Newton’s actor (`actor_tools.py`).
`notes`	`list[dict[str, Any] \| str] \| None`	`None`	Notes (code, message, severity, gate). Written by planner (`service.py`) and Newton’s actor (`actor_tools.py`).
`newton_session`	`str \| None`	`None`	Newton session that drafted this record. Written by planner (`service.py`) and Newton’s actor (`actor_tools.py`).
`reader_findings`	`list[str] \| None`	`None`	Reader finding ids the actor drew on (§17). Written by planner (`service.py`) and Newton’s actor (`actor_tools.py`).
`reader_findings_untrusted`	`list[str] \| None`	`None`	Reader finding ids marked untrusted text (§17). Written by planner (`service.py`) and Newton’s actor (`actor_tools.py`).
`injection_flagged_findings`	`list[str] \| None`	`None`	Reader findings flagged as possible prompt injection (§17). Written by planner (`service.py`) and Newton’s actor (`actor_tools.py`).

Source: packages/ump-spec/src/ump_spec/models/decision.py, line 47

Approval

class

class Approval(StrictModel)

Fields

Name	Type	Default	Description
`artifact_id`	`str`
`artifact_version`	`int`
`approver`	`str`
`submitter`	`str \| None`	`None`
`decision`	`Literal['approve', 'reject']`	`'approve'`
`comment`	`str \| None`	`None`
`payload_digest`	`str`
`signature`	`str`
`key_version`	`str`
`acknowledged_fallbacks`	`list[str]`	`list()`	Fallback codes acknowledged.
`at`	`datetime \| None`	`None`

Source: packages/ump-spec/src/ump_spec/models/decision.py, line 143

Constants

Name	Type	Value	Description
`MaterialityTier`		`Literal['light', 'standard', 'full']`