agent_sdk_eval/
lib.rs

1//! Optional evaluation framework primitives for Agent SDK runs.
2//!
3//! This crate owns post-hoc evaluation contracts over core traces and evidence.
4//! It does not run agents, append journals, publish events, choose evaluator
5//! models, or define product-specific success rubrics.
6
7pub mod comparison;
8pub mod evaluator;
9pub mod evidence;
10pub mod identity;
11pub mod metrics;
12pub mod report;
13pub mod request;
14pub mod scope;
15pub mod testing;
16
17pub use comparison::ComparisonDesign;
18pub use evaluator::Evaluator;
19pub use evidence::{EvidenceBundle, EvidenceItem, EvidenceRole, SupportRefValidation};
20pub use identity::EvaluationId;
21pub use metrics::{ToolTraceMetric, TraceMetrics, TraceMetricsComparison};
22pub use report::{
23    EvaluationConfidence, EvaluationMetricDelta, EvaluationReport, EvaluationVerdict,
24    EvaluatorJudgment,
25};
26pub use request::{EvaluationBudget, EvaluationRequest, EvaluationUsage};
27pub use scope::{
28    EvaluationCriterion, EvaluationScope, EvaluationSubject, EvaluationSubjectRole, ExpectedOutcome,
29};
agent_sdk_eval/lib.rs

agent_sdk_eval/
lib.rs
