pub struct GoldenCase {Show 36 fields
pub query: String,
pub expect: Vec<String>,
pub garbage: bool,
pub context_must: Vec<String>,
pub context_must_not: Vec<String>,
pub context_edges_must: Vec<ContextEdgeExpectation>,
pub route_relation_must: Vec<RelationMatchExpectation>,
pub expected_focus_route: Option<String>,
pub expected_focus_routes: Vec<String>,
pub expected_compound_primary: Option<String>,
pub expected_vocabulary_gap_terms: Vec<String>,
pub expected_no_vocabulary_gap: bool,
pub expected_vocab_editable_targets: Vec<String>,
pub expected_vocab_non_editable_targets: Vec<String>,
pub expected_omitted_context: Vec<ContextOmissionExpectation>,
pub expected_omitted_compound_anchors: Vec<ContextOmissionExpectation>,
pub min_confidence: Option<String>,
pub expected_confidence: Option<String>,
pub expected_partition: Option<String>,
pub combo: Option<String>,
pub tool: Option<String>,
pub target_node: Option<String>,
pub expected_pass: Option<bool>,
pub file: Option<String>,
pub line_start: Option<usize>,
pub line_end: Option<usize>,
pub quote: Option<String>,
pub expected_holds: Option<bool>,
pub class: Option<String>,
pub difficulty: Option<String>,
pub requires_code: bool,
pub requires_docs: bool,
pub expected_operators: Vec<String>,
pub expected_required_checks: Vec<String>,
pub expected_risk_hints: Vec<String>,
pub expected_profile: Option<String>,
}Expand description
One golden judgment. expect is the set of acceptable node ids (any one counts as a
correct top hit). garbage marks a nonsense query that MUST be rejected.
Fields§
§query: String§expect: Vec<String>§garbage: bool§context_must: Vec<String>Node ids that SHOULD appear in the activated context for this query (support recall).
context_must_not: Vec<String>Node ids that should NOT appear in the context (noise — e.g. a spurious mention leak).
context_edges_must: Vec<ContextEdgeExpectation>Directed relation edges that SHOULD appear in the focused context.
route_relation_must: Vec<RelationMatchExpectation>Structured relation evidence that SHOULD justify the top route hit.
expected_focus_route: Option<String>Optional focused-route node id expected after graph-aware route selection.
expected_focus_routes: Vec<String>Optional set of focused-route node ids accepted for graph-aware route selection. Use when a type and the helper that materializes it are both valid anchors for the same agent task.
expected_compound_primary: Option<String>Optional compound primary node id expected for multi-anchor/faceted focus queries.
expected_vocabulary_gap_terms: Vec<String>Vocabulary-gap terms expected from compound focus.
expected_no_vocabulary_gap: boolWhether compound focus is expected to produce no vocabulary gap.
expected_vocab_editable_targets: Vec<String>Editable vocabulary-repair targets expected from compound focus.
expected_vocab_non_editable_targets: Vec<String>Non-editable vocabulary-repair targets expected from compound focus.
expected_omitted_context: Vec<ContextOmissionExpectation>Context candidates that should be reported as considered but omitted from focus.
expected_omitted_compound_anchors: Vec<ContextOmissionExpectation>Compound anchors that should be reported as considered but omitted from the selected multi-anchor interpretation.
min_confidence: Option<String>Optional minimum confidence band required for the top route hit.
expected_confidence: Option<String>Optional exact confidence band expected for the top route hit.
expected_partition: Option<String>Optional generated EKF partition expected for the top/focused route node.
combo: Option<String>Optional package combo evaluator id, e.g. ground_gate or verify_receipt.
tool: Option<String>Optional tool id used by combo evaluators that exercise a concrete MCP/engine tool.
target_node: Option<String>Optional target node for combo evaluators that attach an operation to an existing node.
expected_pass: Option<bool>Expected boolean result for combo evaluators with pass/fail semantics.
file: Option<String>Receipt file for verify_receipt combo cases. Relative paths resolve from package root.
line_start: Option<usize>1-indexed receipt start line for verify_receipt.
line_end: Option<usize>Optional 1-indexed receipt end line for verify_receipt.
quote: Option<String>Expected quote for verify_receipt.
expected_holds: Option<bool>Expected verify_receipt result.
class: Option<String>Optional intent-class tag (capability, exact-lookup, troubleshoot, ambiguous, garbage-*, …) — aggregated per class in the report so a large stratified set shows where it’s weak.
difficulty: Option<String>Optional dogfood difficulty bucket: easy, intermediate, or complex.
requires_code: boolWhether the task should pull at least one code node into focused context.
requires_docs: boolWhether the task should pull at least one docs/skill/section node into focused context.
expected_operators: Vec<String>Operators that should appear in the brief decomposition, serialized as snake_case names.
expected_required_checks: Vec<String>Required checks that should be present in the selected brief/decomposition.
expected_risk_hints: Vec<String>Risk/review hint titles that should be present in the selected brief/decomposition.
expected_profile: Option<String>EKF workflow profile id expected for package-level workflow planning.
Trait Implementations§
Source§impl Clone for GoldenCase
impl Clone for GoldenCase
Source§fn clone(&self) -> GoldenCase
fn clone(&self) -> GoldenCase
1.0.0 (const: unstable) · Source§fn clone_from(&mut self, source: &Self)
fn clone_from(&mut self, source: &Self)
source. Read more