@@ -83,7 +83,7 @@ func effectiveDryRun(cfg Config) bool {
 // (2) score, display, confirm, and purchase.
 func runToolMultiService(ctx context.Context, cfg Config) {
 	if cfg.CSVInput != "" {
-		runToolFromCSV(ctx, cfg)
+		runCSVPathOrFatal(ctx, cfg)
 		return
 	}
 
@@ -272,8 +272,20 @@ func buildServiceStats(recs []common.Recommendation, results []common.PurchaseRe
 	return stats
 }
 
+// runCSVPathOrFatal runs the CSV purchase path and exits fatally on error.
+// It isolates the error-to-fatal glue from runToolMultiService so that path
+// stays under the cyclomatic-complexity budget while runToolFromCSV remains
+// unit-testable via its returned error.
+func runCSVPathOrFatal(ctx context.Context, cfg Config) {
+	if err := runToolFromCSV(ctx, cfg); err != nil {
+		log.Fatalf("%v", err)
+	}
+}
+
 // runToolFromCSV processes recommendations from a CSV input file.
-func runToolFromCSV(ctx context.Context, cfg Config) {
+// It returns an error instead of exiting so the orchestration glue is
+// unit-testable; the caller (runCSVPathOrFatal) turns errors fatal.
+func runToolFromCSV(ctx context.Context, cfg Config) error {
 	isDryRun := effectiveDryRun(cfg)
 	printRunMode(isDryRun)
 
@@ -284,7 +296,7 @@ func runToolFromCSV(ctx context.Context, cfg Config) {
 	// Read recommendations from CSV
 	recs, err := loadRecommendationsFromCSV(cfg.CSVInput)
 	if err != nil {
-		log.Fatalf("Failed to read CSV file: %v", err)
+		return fmt.Errorf("failed to read CSV file: %w", err)
 	}
 
 	AppLogger.Printf("✅ Loaded %d recommendations from CSV\n", len(recs))
@@ -294,7 +306,7 @@ func runToolFromCSV(ctx context.Context, cfg Config) {
 
 	if len(recs) == 0 {
 		AppLogger.Println("⚠️  No recommendations to process after filtering")
-		return
+		return nil
 	}
 
 	// Load AWS configuration
@@ -305,7 +317,7 @@ func runToolFromCSV(ctx context.Context, cfg Config) {
 	}
 	awsCfg, err := awsconfig.LoadDefaultConfig(ctx, configOptions...)
 	if err != nil {
-		log.Fatalf("Failed to load AWS config: %v", err)
+		return fmt.Errorf("failed to load AWS config: %w", err)
 	}
 
 	// Create account alias cache for lookup
@@ -386,6 +398,7 @@ func runToolFromCSV(ctx context.Context, cfg Config) {
 	// Print final summary using the post-dedup slice so counts match what was
 	// actually processed, not the pre-dedup input passed into the outer loop.
 	printMultiServiceSummary(allAdjustedRecs, allResults, serviceStats, isDryRun)
+	return nil
 }
 
 // filterAndAdjustRecommendations applies filters, coverage, count override, and instance limits to recommendations.
 
@@ -9,7 +9,10 @@ import (
 
 	"github.com/LeanerCloud/CUDly/pkg/common"
 	"github.com/aws/aws-sdk-go-v2/aws"
+	awsrds "github.com/aws/aws-sdk-go-v2/service/rds"
+	rdstypes "github.com/aws/aws-sdk-go-v2/service/rds/types"
 	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
 )
 
 // ==================== Tests for coverage improvement ====================
@@ -18,72 +21,124 @@ import (
 
 // ==================== Tests for queryMajorEngineVersions ====================
 
-func TestQueryMajorEngineVersions_Success(t *testing.T) {
-	// This test verifies the logic of queryMajorEngineVersions without mocking
-	// Since it requires AWS credentials, we test the error cases and structure
+// engineKeyedRDSMajorVersionsStub implements RDSMajorVersionsClient with
+// canned per-engine responses and records which engines were queried, so
+// queryMajorEngineVersionsWithClient can be tested hermetically (no AWS
+// credentials, no network).
+type engineKeyedRDSMajorVersionsStub struct {
+	versionsByEngine map[string][]rdstypes.DBMajorEngineVersion
+	errByEngine      map[string]error
+	enginesQueried   []string
+}
 
-	ctx := context.Background()
-	origCfg := toolCfg
-	defer func() { toolCfg = origCfg }()
+func (s *engineKeyedRDSMajorVersionsStub) DescribeDBMajorEngineVersions(
+	_ context.Context,
+	params *awsrds.DescribeDBMajorEngineVersionsInput,
+	_ ...func(*awsrds.Options),
+) (*awsrds.DescribeDBMajorEngineVersionsOutput, error) {
+	engine := aws.ToString(params.Engine)
+	s.enginesQueried = append(s.enginesQueried, engine)
+	if err := s.errByEngine[engine]; err != nil {
+		return nil, err
+	}
+	return &awsrds.DescribeDBMajorEngineVersionsOutput{
+		DBMajorEngineVersions: s.versionsByEngine[engine],
+	}, nil
+}
 
-	// Test with empty profile (should work if AWS credentials are configured)
-	toolCfg.Profile = ""
-	toolCfg.ValidationProfile = ""
+// TestQueryMajorEngineVersionsWithClient_Success replaces a former
+// live-credential test: it stubs the RDS client and asserts the merged
+// engine:major-version map deterministically.
+func TestQueryMajorEngineVersionsWithClient_Success(t *testing.T) {
+	stub := &engineKeyedRDSMajorVersionsStub{
+		versionsByEngine: map[string][]rdstypes.DBMajorEngineVersion{
+			"mysql":    {rdsMajorVersion("mysql", "5.7"), rdsMajorVersion("mysql", "8.0")},
+			"postgres": {rdsMajorVersion("postgres", "13")},
+		},
+	}
 
-	// This will attempt to load AWS config - may fail without credentials
-	result, err := queryMajorEngineVersions(ctx, toolCfg)
+	result, err := queryMajorEngineVersionsWithClient(context.Background(), stub)
+	require.NoError(t, err)
+
+	assert.ElementsMatch(t,
+		[]string{"mysql", "postgres", "aurora-mysql", "aurora-postgresql"},
+		stub.enginesQueried,
+		"must query every supported engine exactly once")
+
+	require.Len(t, result, 3)
+	assert.Equal(t, "mysql", result["mysql:5.7"].Engine)
+	assert.Equal(t, "5.7", result["mysql:5.7"].MajorEngineVersion)
+	assert.Equal(t, "8.0", result["mysql:8.0"].MajorEngineVersion)
+	assert.Equal(t, "13", result["postgres:13"].MajorEngineVersion)
+	require.Len(t, result["postgres:13"].SupportedEngineLifecycles, 1,
+		"lifecycle data must be carried through from the API response")
+}
 
-	// Either succeeds with valid credentials or fails gracefully
-	if err != nil {
-		assert.Error(t, err)
-		assert.Contains(t, err.Error(), "failed to load AWS config")
-	} else {
-		// If it succeeds, verify the result structure
-		assert.NotNil(t, result)
-		// Result can be empty if no versions are found
+// TestQueryMajorEngineVersionsWithClient_EngineErrorContinues asserts the
+// warn-and-continue contract: one engine failing must not drop the results
+// of the others, and the overall call still succeeds.
+func TestQueryMajorEngineVersionsWithClient_EngineErrorContinues(t *testing.T) {
+	stub := &engineKeyedRDSMajorVersionsStub{
+		versionsByEngine: map[string][]rdstypes.DBMajorEngineVersion{
+			"postgres": {rdsMajorVersion("postgres", "15")},
+		},
+		errByEngine: map[string]error{
+			"mysql": errors.New("DescribeDBMajorEngineVersions throttled"),
+		},
 	}
+
+	result, err := queryMajorEngineVersionsWithClient(context.Background(), stub)
+	require.NoError(t, err, "per-engine API failures are warn-and-continue")
+
+	assert.ElementsMatch(t,
+		[]string{"mysql", "postgres", "aurora-mysql", "aurora-postgresql"},
+		stub.enginesQueried,
+		"a failing engine must not stop the remaining engine queries")
+	require.Len(t, result, 1)
+	assert.Equal(t, "15", result["postgres:15"].MajorEngineVersion)
 }
 
-func TestQueryMajorEngineVersions_ProfileHandling(t *testing.T) {
+// TestQueryMajorEngineVersions_ProfileSelection asserts validation-profile
+// precedence deterministically: the configured profiles point at names
+// guaranteed not to exist in the (redirected, empty) shared config, so
+// config loading fails fast without any live AWS call and the error names
+// the profile the function actually selected.
+func TestQueryMajorEngineVersions_ProfileSelection(t *testing.T) {
 	ctx := context.Background()
-	origCfg := toolCfg
-	defer func() { toolCfg = origCfg }()
 
 	tests := []struct {
 		name              string
 		profile           string
 		validationProfile string
+		wantProfileInErr  string
 	}{
 		{
-			name:              "Uses validation profile if set",
-			profile:           "main-profile",
-			validationProfile: "validation-profile",
+			name:              "validation profile takes precedence",
+			profile:           "cudly-test-main-profile-does-not-exist",
+			validationProfile: "cudly-test-validation-profile-does-not-exist",
+			wantProfileInErr:  "cudly-test-validation-profile-does-not-exist",
 		},
 		{
-			name:              "Falls back to main profile",
-			profile:           "main-profile",
-			validationProfile: "",
-		},
-		{
-			name:              "Empty profiles use default",
-			profile:           "",
-			validationProfile: "",
+			name:             "falls back to main profile",
+			profile:          "cudly-test-main-profile-does-not-exist",
+			wantProfileInErr: "cudly-test-main-profile-does-not-exist",
 		},
 	}
 
 	for _, tt := range tests {
 		t.Run(tt.name, func(t *testing.T) {
-			toolCfg.Profile = tt.profile
-			toolCfg.ValidationProfile = tt.validationProfile
-
-			// This will attempt to load config - may fail without valid profiles
-			_, err := queryMajorEngineVersions(ctx, toolCfg)
-
-			// We just verify it doesn't panic - actual AWS calls may fail
-			// Error is acceptable for invalid profiles
-			if err != nil {
-				assert.Error(t, err)
-			}
+			t.Setenv("AWS_CONFIG_FILE", os.DevNull)
+			t.Setenv("AWS_SHARED_CREDENTIALS_FILE", os.DevNull)
+
+			_, err := queryMajorEngineVersions(ctx, Config{
+				Profile:           tt.profile,
+				ValidationProfile: tt.validationProfile,
+			})
+
+			require.Error(t, err)
+			assert.Contains(t, err.Error(), "failed to load AWS config")
+			assert.Contains(t, err.Error(), tt.wantProfileInErr,
+				"error must name the selected profile")
 		})
 	}
 }