pub struct ExperimentEvidenceBundle {Show 36 fields
pub bundle_id: String,
pub candidate_id: String,
pub eval_id: String,
pub version_id: String,
pub supersedes_claim_version_id: Option<ClaimVersionId>,
pub relation_lineage_hints: RelationLineageHints,
pub scores: ScoreVector,
pub hypotheses: Vec<CausalHypothesis>,
pub verification: Option<VerificationPlan>,
pub trace_id: Option<String>,
pub experiment_diff: Option<ExperimentDiff>,
pub attribution_json: Option<String>,
pub assessment: Option<EvidenceAssessment>,
pub warnings: Vec<String>,
pub created_at: String,
pub run_id: Option<String>,
pub attempt_id: Option<String>,
pub causal_question: Option<String>,
pub unit_definition: Option<String>,
pub bundle_scope: Option<BundleScope>,
pub pair_comparability: Option<PairComparability>,
pub claim_strength: ClaimStrength,
pub identification_rationale: Option<String>,
pub known_threats: Vec<String>,
pub patch_hash: Option<String>,
pub treatment: Option<Treatment>,
pub outcome: Option<String>,
pub covariates: Option<Covariates>,
pub promotion_state: Option<PromotionState>,
pub primary_effect: Option<TypedLocatedEffect>,
pub all_effects: Vec<TypedLocatedEffect>,
pub hypothesis_edges: Vec<HypothesisEdge>,
pub receipts: Vec<ReceiptRef>,
pub verification_trials: Vec<VerificationTrial>,
pub refutation_artifacts: Vec<RefutationArtifact>,
pub sealed: bool,
}Expand description
A bundle of evidence collected during a forge evaluation run.
One ExperimentEvidenceBundle = one paired experiment on one fixed workload slice.
One bundle may contain multiple effects but names exactly one primary effect.
Claim strength for Phase 5 is always ProvisionalSinglePair.
semantic_memory_forge::EvidenceBundle is the authoritative cross-crate
evidence contract. This local type is an authoring wrapper that carries
forge-engine-only working fields and round-trips them explicitly through the
canonical bundle metadata under [LOCAL_AUTHORING_METADATA_KEY].
Fields§
§bundle_id: StringUnique identifier for this bundle.
candidate_id: StringThe candidate that produced this evidence.
eval_id: StringEval run that generated the evidence.
version_id: StringVersion of the algebra spec used.
supersedes_claim_version_id: Option<ClaimVersionId>Real prior claim-lineage version when known.
This is optional and carried through to the export envelope so that downstream projection imports can preserve version-aware supersession.
relation_lineage_hints: RelationLineageHintsExplicit relation-version lineage hints for exportable relations, when known.
These hints are advisory only. They do not mint lineage, and export leaves
supersedes_relation_version_id unset when no exact hint is present.
scores: ScoreVectorComputed scores.
hypotheses: Vec<CausalHypothesis>Causal hypotheses derived from this run (legacy field, prefer hypothesis_edges).
verification: Option<VerificationPlan>Verification plan for follow-up validation.
trace_id: Option<String>Trace ID for cross-crate correlation.
experiment_diff: Option<ExperimentDiff>Typed experiment diff (baseline vs patched), if from a paired experiment.
attribution_json: Option<String>Attribution result from CEA, if available.
assessment: Option<EvidenceAssessment>Deterministic evidence assessment.
warnings: Vec<String>Warnings generated during evidence collection.
created_at: StringWhen this bundle was recorded by Forge.
run_id: Option<String>Forge domain ID for the experiment run.
attempt_id: Option<String>Forge domain ID for the patch attempt.
causal_question: Option<String>The causal question this bundle answers.
unit_definition: Option<String>Explicit description of the unit of observation.
bundle_scope: Option<BundleScope>Scope of this bundle.
pair_comparability: Option<PairComparability>Explicit paired comparability verdict when the source captured it.
claim_strength: ClaimStrengthHow strong a claim this bundle makes.
identification_rationale: Option<String>Why we think confounding is controlled.
known_threats: Vec<String>Known threats to validity.
patch_hash: Option<String>Content hash of the patch.
treatment: Option<Treatment>The treatment applied.
outcome: Option<String>Outcome summary.
covariates: Option<Covariates>Measured covariates.
promotion_state: Option<PromotionState>Promotion state for the claim represented by this bundle, when known.
primary_effect: Option<TypedLocatedEffect>The primary effect attributed to the patch.
all_effects: Vec<TypedLocatedEffect>All effects observed in the experiment.
hypothesis_edges: Vec<HypothesisEdge>Typed hypothesis edges (Phase 5 edge model).
receipts: Vec<ReceiptRef>Receipts proving artifact integrity.
verification_trials: Vec<VerificationTrial>Canonical per-side verification trials for this causal claim.
refutation_artifacts: Vec<RefutationArtifact>Refutation artifacts generated by verification infrastructure.
sealed: boolWhether this bundle is sealed (immutable after creation).
Implementations§
Source§impl ExperimentEvidenceBundle
impl ExperimentEvidenceBundle
Sourcepub fn seal(&mut self) -> ForgeResult<()>
pub fn seal(&mut self) -> ForgeResult<()>
Seal the bundle, making it logically immutable. Returns an error if the bundle is already sealed.
Sourcepub fn check_not_sealed(&self) -> ForgeResult<()>
pub fn check_not_sealed(&self) -> ForgeResult<()>
Returns an error if the bundle is sealed and mutation is attempted.
Sourcepub fn to_canonical_evidence_bundle(&self) -> EvidenceBundle
pub fn to_canonical_evidence_bundle(&self) -> EvidenceBundle
Project this local bundle into the canonical Forge evidence contract.
The canonical semantic_memory_forge::EvidenceBundle remains the
authoritative cross-crate contract. Local forge-engine-only authoring
fields are preserved explicitly inside canonical metadata rather than
becoming a second drifting wire contract.
Sourcepub fn from_canonical_evidence_bundle(
canonical: &EvidenceBundle,
) -> ForgeResult<Self>
pub fn from_canonical_evidence_bundle( canonical: &EvidenceBundle, ) -> ForgeResult<Self>
Rebuild the local authoring wrapper from the canonical Forge evidence contract.
The canonical bundle is authoritative. When local authoring metadata is present, it is used only to recover forge-engine-specific working fields that are intentionally outside the canonical contract.
Sourcepub fn render_causal_question(
patch_summary: &str,
outcome_summary: &str,
workload_id: &str,
scope_summary: &str,
) -> String
pub fn render_causal_question( patch_summary: &str, outcome_summary: &str, workload_id: &str, scope_summary: &str, ) -> String
Render the causal question for this bundle.
Sourcepub fn superseded_effect_relation_version_id(
&self,
source: EffectRelationLineageSource,
effect: &TypedLocatedEffect,
) -> Option<RelationVersionId>
pub fn superseded_effect_relation_version_id( &self, source: EffectRelationLineageSource, effect: &TypedLocatedEffect, ) -> Option<RelationVersionId>
Return the prior relation version for an effect-derived export relation, when known.
Sourcepub fn superseded_hypothesis_relation_version_id(
&self,
edge: &HypothesisEdge,
) -> Option<RelationVersionId>
pub fn superseded_hypothesis_relation_version_id( &self, edge: &HypothesisEdge, ) -> Option<RelationVersionId>
Return the prior relation version for a hypothesis-edge export relation, when known.
Sourcepub fn superseded_verification_trial_relation_version_id(
&self,
trial: &VerificationTrial,
) -> Option<RelationVersionId>
pub fn superseded_verification_trial_relation_version_id( &self, trial: &VerificationTrial, ) -> Option<RelationVersionId>
Return the prior relation version for a verification-trial export relation, when known.
Sourcepub fn superseded_refutation_relation_version_id(
&self,
artifact: &RefutationArtifact,
) -> Option<RelationVersionId>
pub fn superseded_refutation_relation_version_id( &self, artifact: &RefutationArtifact, ) -> Option<RelationVersionId>
Return the prior relation version for a refutation-artifact export relation, when known.
Sourcepub fn to_episode_meta(&self) -> Value
pub fn to_episode_meta(&self) -> Value
Convert the bundle into episode metadata suitable for semantic-memory storage.
Sourcepub fn to_episode_content(&self) -> String
pub fn to_episode_content(&self) -> String
Convert the bundle into episode content text for embedding.
Includes: patch summary, primary causal question, primary effect summary, key covariates, claim strength, and verification recommendations. Does NOT include: giant raw logs, unstable formatting, or missing identifiers.
Trait Implementations§
Source§impl Clone for ExperimentEvidenceBundle
impl Clone for ExperimentEvidenceBundle
Source§fn clone(&self) -> ExperimentEvidenceBundle
fn clone(&self) -> ExperimentEvidenceBundle
1.0.0 (const: unstable) · Source§fn clone_from(&mut self, source: &Self)
fn clone_from(&mut self, source: &Self)
source. Read more