Function TestSaveRunSessionsJSON

pkg/evaluation/save_test.go:121–291

(t *testing.T)

}

func TestSaveRunSessionsJSON(t *testing.T) {
	t.Parallel()

	outputDir := t.TempDir()

	// Create sessions with different content
	sess1 := session.New(
		session.WithTitle("eval-json-1"),
		session.WithUserMessage("What is the capital of France?"),
	)
	sess1.InputTokens = 100
	sess1.OutputTokens = 50
	sess1.Cost = 0.01
	sess1.Evals = &session.EvalCriteria{
		Relevance: []string{"mentions Paris", "mentions France"},
	}

	sess2 := session.New(
		session.WithTitle("eval-json-2"),
		session.WithUserMessage("What is 2+2?"),
	)
	sess2.InputTokens = 80
	sess2.OutputTokens = 30
	sess2.Cost = 0.005
	sess2.Evals = &session.EvalCriteria{
		Relevance: []string{"gives the correct answer", "explains the math"},
	}

	// Create an eval run with sessions and eval criteria
	run := &EvalRun{
		Name:      "test-json-001",
		Timestamp: time.Now(),
		Duration:  42 * time.Second,
		Config: Config{
			AgentFilename: "./test-agent.yaml",
			JudgeModel:    "anthropic/claude-opus-4-5",
			Concurrency:   4,
			EvalsDir:      "./evals",
		},
		Summary: Summary{
			TotalEvals:  3,
			FailedEvals: 1,
			TotalCost:   0.015,
		},
		Results: []Result{
			{
				Title:             "eval-json-1",
				Question:          "What is the capital of France?",
				Response:          "Paris is the capital of France.",
				Cost:              0.01,
				OutputTokens:      50,
				RelevancePassed:   2,
				RelevanceExpected: 2,
				RelevanceResults: []RelevanceResult{
					{Criterion: "mentions Paris", Passed: true, Reason: "response includes Paris"},
					{Criterion: "mentions France", Passed: true, Reason: "response includes France"},
				},
				Session: sess1,