swink-agent-eval 0.8.1

Evaluation framework for swink-agent: trajectory tracing, golden path verification, and cost governance
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82

#![forbid(unsafe_code)]
//! Evaluation framework for swink-agent.
//!
//! Provides trajectory tracing, golden path verification, response matching,
//! and cost/latency governance for LLM-powered agent loops.
//!
//! # Quick Start
//!
//! ```rust,ignore
//! use swink_agent_eval::{EvalRunner, EvalSet, EvalCase};
//!
//! let set = EvalSet { id: "demo".into(), name: "Demo".into(), description: None, cases: vec![] };
//! let runner = EvalRunner::with_defaults();
//! let result = runner.run_set(&set, &my_factory).await?;
//! println!("Passed: {}/{}", result.summary.passed, result.summary.total_cases);
//! ```

pub mod aggregator;
mod audit;
mod budget;
pub mod cache;
mod efficiency;
mod environment_state;
mod error;
mod evaluator;
pub mod evaluators;
mod gate;
#[cfg(feature = "generation")]
pub mod generation;
pub mod judge;
mod match_;
#[cfg(feature = "judge-core")]
pub mod prompt;
pub mod report;
mod response;
mod runner;
mod score;
mod semantic_tool_parameter;
mod semantic_tool_selection;
#[cfg(feature = "simulation")]
pub mod simulation;
mod store;
#[cfg(feature = "telemetry")]
pub mod telemetry;
pub mod testing;
#[cfg(feature = "trace-ingest")]
pub mod trace;
mod trajectory;
mod types;
mod url_filter;
#[cfg(feature = "yaml")]
mod yaml;

// ─── Public API ─────────────────────────────────────────────────────────────

pub use aggregator::{Aggregator, AllPass, AnyPass, Average, Weighted};
pub use audit::AuditedInvocation;
pub use budget::BudgetEvaluator;
pub use efficiency::EfficiencyEvaluator;
pub use environment_state::EnvironmentStateEvaluator;
pub use error::EvalError;
pub use evaluator::{Evaluator, EvaluatorRegistry};
pub use gate::{GateConfig, GateResult, check_gate};
pub use judge::{JudgeClient, JudgeError, JudgeVerdict};
pub use match_::{MatchMode, TrajectoryMatcher};
pub use response::ResponseMatcher;
pub use runner::{AgentFactory, EvalRunner};
pub use score::{Score, Verdict};
pub use semantic_tool_parameter::SemanticToolParameterEvaluator;
pub use semantic_tool_selection::SemanticToolSelectionEvaluator;
pub use store::{EvalStore, FsEvalStore};
pub use testing::{MockJudge, PanickingMockJudge, SlowMockJudge};
pub use trajectory::TrajectoryCollector;
pub use types::{
    BudgetConstraints, CASE_NAMESPACE, CaseFingerprint, EnvironmentState, EvalCase, EvalCaseResult,
    EvalMetricResult, EvalSet, EvalSetResult, EvalSummary, ExpectedToolCall, Invocation,
    RecordedToolCall, ResponseCriteria, StateCapture, ToolIntent, TurnRecord, validate_eval_case,
    validate_eval_set,
};
pub use url_filter::{DefaultUrlFilter, UrlFilter};
#[cfg(feature = "yaml")]
pub use yaml::load_eval_set_yaml;