truthound-dashboard 1.4.4__py3-none-any.whl → 1.5.1__py3-none-any.whl

truthound_dashboard/core/truthound_adapter.py

@@ -6,18 +6,41 @@ enabling non-blocking validation operations in the FastAPI application.
 The adapter uses ThreadPoolExecutor to run synchronous truthound
 functions without blocking the async event loop.
 
+Architecture:
+    Dashboard Services
+           ↓
+    TruthoundAdapter (this module)
+           ↓
+    truthound library (external)
+
+The adapter is designed for loose coupling with truthound:
+- Protocol-based interfaces for type checking
+- Graceful fallbacks when truthound versions differ
+- All truthound interactions are isolated in this module
+
 Features:
-- Async wrappers for all truthound functions
+- Async wrappers for all truthound functions (check, learn, profile, compare, scan, mask)
+- Support for both file paths and DataSource objects
 - Automatic sampling for large datasets (100MB+ files)
+- ValidationResult conversion for reporter integration
 - Configurable sample size and sampling methods
 
 Example:
     adapter = get_adapter()
+
+    # With file path
     result = await adapter.check("/path/to/data.csv")
-    schema = await adapter.learn("/path/to/data.csv")
+
+    # With DataSource
+    from truthound_dashboard.core.datasource_factory import create_datasource
+    source = create_datasource({"type": "postgresql", "table": "users", ...})
+    result = await adapter.check(source)
 
     # With auto-sampling for large files
     result = await adapter.check_with_sampling("/path/to/large.csv")
+
+    # Convert to ValidationResult for reporters
+    validation_result = result.to_validation_result()
 """
 
 from __future__ import annotations
@@ -28,12 +51,18 @@ from concurrent.futures import ThreadPoolExecutor
 from dataclasses import dataclass
 from functools import partial
 from pathlib import Path
-from typing import Any, Protocol, runtime_checkable
+from typing import TYPE_CHECKING, Any, Protocol, Union, runtime_checkable
 
 import yaml
 
+if TYPE_CHECKING:
+    from truthound_dashboard.core.datasource_factory import SourceConfig
+
 logger = logging.getLogger(__name__)
 
+# Type alias for data input - can be path string or DataSource object
+DataInput = Union[str, Any]
+
 
 @runtime_checkable
 class TruthoundResult(Protocol):
@@ -47,6 +76,9 @@ class TruthoundResult(Protocol):
 class CheckResult:
     """Validation check result.
 
+    This class wraps truthound's Report/ValidationResult and provides
+    a consistent interface for the dashboard regardless of truthound version.
+
     Attributes:
         passed: Whether validation passed (no issues).
         has_critical: Whether critical issues were found.
@@ -56,10 +88,13 @@ class CheckResult:
         high_issues: Number of high severity issues.
         medium_issues: Number of medium severity issues.
         low_issues: Number of low severity issues.
-        source: Data source path.
+        source: Data source path or name.
         row_count: Number of rows validated.
         column_count: Number of columns.
         issues: List of validation issues.
+        run_id: Optional run identifier for tracking.
+        run_time: Optional timestamp of the validation run.
+        _raw_result: Internal reference to the original truthound result.
     """
 
     passed: bool
@@ -74,10 +109,13 @@ class CheckResult:
     row_count: int
     column_count: int
     issues: list[dict[str, Any]]
+    run_id: str | None = None
+    run_time: Any = None
+    _raw_result: Any = None
 
     def to_dict(self) -> dict[str, Any]:
         """Convert to dictionary."""
-        return {
+        result = {
             "passed": self.passed,
             "has_critical": self.has_critical,
             "has_high": self.has_high,
@@ -91,6 +129,39 @@ class CheckResult:
             "column_count": self.column_count,
             "issues": self.issues,
         }
+        if self.run_id:
+            result["run_id"] = self.run_id
+        if self.run_time:
+            result["run_time"] = (
+                self.run_time.isoformat()
+                if hasattr(self.run_time, "isoformat")
+                else str(self.run_time)
+            )
+        return result
+
+    def to_validation_result(self) -> Any:
+        """Convert to truthound's ValidationResult format for reporters.
+
+        This enables using truthound's reporters directly with this result.
+
+        Returns:
+            An object that implements the ValidationResult interface expected
+            by truthound reporters, or the raw result if available.
+        """
+        # If we have the raw truthound result, prefer using it
+        if self._raw_result is not None:
+            # Check if it's already a ValidationResult
+            if hasattr(self._raw_result, "results") and hasattr(
+                self._raw_result, "run_id"
+            ):
+                return self._raw_result
+            # It's a Report - try to convert
+            return self._create_validation_result_mock()
+        return self._create_validation_result_mock()
+
+    def _create_validation_result_mock(self) -> "_ValidationResultMock":
+        """Create a mock ValidationResult for reporter compatibility."""
+        return _ValidationResultMock(self)
 
 
 @dataclass
@@ -122,32 +193,190 @@ class LearnResult:
         }
 
 
+@dataclass
+class ColumnProfileResult:
+    """Column-level profile result matching truthound's ColumnProfile structure.
+
+    Attributes:
+        name: Column name.
+        physical_type: Polars data type (string).
+        inferred_type: Inferred logical type (e.g., email, phone, integer).
+        row_count: Number of rows.
+        null_count: Number of null values.
+        null_ratio: Ratio of null values (0.0-1.0).
+        empty_string_count: Number of empty strings.
+        distinct_count: Number of distinct values.
+        unique_ratio: Ratio of unique values (0.0-1.0).
+        is_unique: Whether all values are unique.
+        is_constant: Whether all values are the same.
+        distribution: Statistical distribution (for numeric columns).
+        top_values: Most frequent values.
+        bottom_values: Least frequent values.
+        min_length: Minimum string length (for string columns).
+        max_length: Maximum string length (for string columns).
+        avg_length: Average string length (for string columns).
+        detected_patterns: Detected patterns (for string columns).
+        min_date: Minimum date (for datetime columns).
+        max_date: Maximum date (for datetime columns).
+        date_gaps: Number of date gaps (for datetime columns).
+        suggested_validators: List of suggested validator names.
+        profile_duration_ms: Time taken to profile this column.
+    """
+
+    name: str
+    physical_type: str
+    inferred_type: str = "unknown"
+    row_count: int = 0
+    null_count: int = 0
+    null_ratio: float = 0.0
+    empty_string_count: int = 0
+    distinct_count: int = 0
+    unique_ratio: float = 0.0
+    is_unique: bool = False
+    is_constant: bool = False
+    distribution: dict[str, Any] | None = None
+    top_values: list[dict[str, Any]] | None = None
+    bottom_values: list[dict[str, Any]] | None = None
+    min_length: int | None = None
+    max_length: int | None = None
+    avg_length: float | None = None
+    detected_patterns: list[dict[str, Any]] | None = None
+    min_date: str | None = None
+    max_date: str | None = None
+    date_gaps: int = 0
+    suggested_validators: list[str] | None = None
+    profile_duration_ms: float = 0.0
+
+    def to_dict(self) -> dict[str, Any]:
+        """Convert to dictionary."""
+        result = {
+            "name": self.name,
+            "physical_type": self.physical_type,
+            "inferred_type": self.inferred_type,
+            "row_count": self.row_count,
+            "null_count": self.null_count,
+            "null_ratio": self.null_ratio,
+            "empty_string_count": self.empty_string_count,
+            "distinct_count": self.distinct_count,
+            "unique_ratio": self.unique_ratio,
+            "is_unique": self.is_unique,
+            "is_constant": self.is_constant,
+            "profile_duration_ms": self.profile_duration_ms,
+        }
+        if self.distribution:
+            result["distribution"] = self.distribution
+        if self.top_values:
+            result["top_values"] = self.top_values
+        if self.bottom_values:
+            result["bottom_values"] = self.bottom_values
+        if self.min_length is not None:
+            result["min_length"] = self.min_length
+            result["max_length"] = self.max_length
+            result["avg_length"] = self.avg_length
+        if self.detected_patterns:
+            result["detected_patterns"] = self.detected_patterns
+        if self.min_date:
+            result["min_date"] = self.min_date
+            result["max_date"] = self.max_date
+            result["date_gaps"] = self.date_gaps
+        if self.suggested_validators:
+            result["suggested_validators"] = self.suggested_validators
+        return result
+
+
 @dataclass
 class ProfileResult:
-    """Data profiling result.
+    """Data profiling result matching truthound's TableProfile structure.
 
     Attributes:
-        source: Data source path.
+        name: Table/source name.
+        source: Data source path or name.
         row_count: Number of rows.
         column_count: Number of columns.
-        size_bytes: Data size in bytes.
-        columns: List of column profile dictionaries.
+        estimated_memory_bytes: Estimated memory usage in bytes.
+        columns: List of column profile results.
+        duplicate_row_count: Number of duplicate rows.
+        duplicate_row_ratio: Ratio of duplicate rows.
+        correlations: Column correlation pairs with coefficients.
+        profiled_at: Timestamp when profile was created.
+        profile_duration_ms: Total profiling duration in milliseconds.
+        size_bytes: Data size in bytes (backward compatibility).
     """
 
+    name: str
     source: str
     row_count: int
     column_count: int
-    size_bytes: int
-    columns: list[dict[str, Any]]
+    estimated_memory_bytes: int
+    columns: list[ColumnProfileResult]
+    duplicate_row_count: int = 0
+    duplicate_row_ratio: float = 0.0
+    correlations: list[tuple[str, str, float]] | None = None
+    profiled_at: str | None = None
+    profile_duration_ms: float = 0.0
+    size_bytes: int = 0  # Backward compatibility
 
     def to_dict(self) -> dict[str, Any]:
         """Convert to dictionary."""
         return {
+            "name": self.name,
             "source": self.source,
             "row_count": self.row_count,
             "column_count": self.column_count,
-            "size_bytes": self.size_bytes,
-            "columns": self.columns,
+            "estimated_memory_bytes": self.estimated_memory_bytes,
+            "size_bytes": self.size_bytes or self.estimated_memory_bytes,
+            "duplicate_row_count": self.duplicate_row_count,
+            "duplicate_row_ratio": self.duplicate_row_ratio,
+            "correlations": self.correlations,
+            "profiled_at": self.profiled_at,
+            "profile_duration_ms": self.profile_duration_ms,
+            "columns": [col.to_dict() for col in self.columns],
+        }
+
+    def get_column(self, name: str) -> ColumnProfileResult | None:
+        """Get column profile by name."""
+        for col in self.columns:
+            if col.name == name:
+                return col
+        return None
+
+    @property
+    def column_names(self) -> list[str]:
+        """Get list of column names."""
+        return [col.name for col in self.columns]
+
+
+@dataclass
+class GenerateSuiteResult:
+    """Validation suite generation result.
+
+    Result from generating validation rules based on profile data.
+
+    Attributes:
+        rules: List of generated validation rules.
+        rule_count: Total number of rules generated.
+        categories: Categories of rules generated.
+        strictness: Strictness level used for generation.
+        yaml_content: Generated rules as YAML string.
+        json_content: Generated rules as JSON-serializable dict.
+    """
+
+    rules: list[dict[str, Any]]
+    rule_count: int
+    categories: list[str]
+    strictness: str
+    yaml_content: str
+    json_content: dict[str, Any]
+
+    def to_dict(self) -> dict[str, Any]:
+        """Convert to dictionary."""
+        return {
+            "rules": self.rules,
+            "rule_count": self.rule_count,
+            "categories": self.categories,
+            "strictness": self.strictness,
+            "yaml_content": self.yaml_content,
+            "json_content": self.json_content,
         }
 
 
@@ -156,8 +385,8 @@ class CompareResult:
     """Drift comparison result.
 
     Attributes:
-        baseline_source: Baseline data source path.
-        current_source: Current data source path.
+        baseline_source: Baseline data source path or name.
+        current_source: Current data source path or name.
         baseline_rows: Number of rows in baseline.
         current_rows: Number of rows in current.
         has_drift: Whether drift was detected.
@@ -197,7 +426,7 @@ class ScanResult:
     """PII scan result.
 
     Attributes:
-        source: Data source path.
+        source: Data source path or name.
         row_count: Number of rows scanned.
         column_count: Number of columns.
         total_columns_scanned: Total columns that were scanned.
@@ -241,7 +470,7 @@ class MaskResult:
     """Data masking result.
 
     Attributes:
-        source: Original data source path.
+        source: Original data source path or name.
         output_path: Path to the masked output file.
         row_count: Number of rows in the masked data.
         column_count: Number of columns in the masked data.
@@ -271,12 +500,30 @@ class MaskResult:
         }
 
 
+def _get_source_name(data: DataInput) -> str:
+    """Get source name from data input.
+
+    Args:
+        data: File path string or DataSource object.
+
+    Returns:
+        Source name string.
+    """
+    if isinstance(data, str):
+        return data
+    # DataSource objects have a name property
+    return getattr(data, "name", str(type(data).__name__))
+
+
 class TruthoundAdapter:
     """Async wrapper for truthound functions.
 
     This adapter provides an async interface to truthound operations,
     running them in a thread pool to avoid blocking the event loop.
 
+    The adapter supports both file paths and DataSource objects for
+    validation, profiling, and other operations.
+
     Attributes:
         max_workers: Maximum number of worker threads.
     """
@@ -292,15 +539,13 @@ class TruthoundAdapter:
 
     async def check(
         self,
-        data: str,
+        data: DataInput,
         *,
         validators: list[str] | None = None,
-        validator_params: dict[str, dict[str, Any]] | None = None,
+        validator_config: dict[str, dict[str, Any]] | None = None,
         schema: str | None = None,
         auto_schema: bool = False,
-        columns: list[str] | None = None,
         min_severity: str | None = None,
-        strict: bool = False,
         parallel: bool = False,
         max_workers: int | None = None,
         pushdown: bool | None = None,
@@ -311,16 +556,17 @@ class TruthoundAdapter:
         All parameters map directly to th.check() for maximum flexibility.
 
         Args:
-            data: Data source path (CSV, Parquet, etc.).
+            data: Data source - can be:
+                - File path string (CSV, Parquet, JSON, etc.)
+                - DataSource object (SQL, Cloud DW, etc.)
             validators: Optional list of validator names to run.
-            validator_params: Optional dict of per-validator parameters.
+            validator_config: Optional dict of per-validator configuration.
                 Format: {"ValidatorName": {"param1": value1, "param2": value2}}
-                Example: {"Null": {"columns": ["a", "b"], "mostly": 0.95}}
+                Example: {"Null": {"columns": ("a", "b"), "mostly": 0.95}}
+                Note: In truthound 2.x, columns should be tuples, not lists.
             schema: Optional path to schema YAML file.
             auto_schema: If True, auto-learns schema for validation.
-            columns: Columns to validate. If None, validates all columns.
             min_severity: Minimum severity to report ("low", "medium", "high", "critical").
-            strict: If True, raises exception on validation failures.
             parallel: If True, uses DAG-based parallel execution.
             max_workers: Max threads for parallel execution.
             pushdown: Enable query pushdown for SQL sources. None uses auto-detection.
@@ -331,36 +577,38 @@ class TruthoundAdapter:
         Raises:
             ImportError: If truthound is not installed.
             FileNotFoundError: If data file doesn't exist.
-            ValidationError: If strict=True and validation fails.
         """
         import truthound as th
 
         # Build kwargs dynamically to avoid passing None for optional params
-        # This ensures truthound uses its own defaults when params are not specified
-        kwargs: dict[str, Any] = {
-            "validators": validators,
-            "schema": schema,
-            "auto_schema": auto_schema,
-            "parallel": parallel,
-        }
+        # Use 'source' parameter for DataSource objects (truthound 2.x API)
+        if isinstance(data, str):
+            kwargs: dict[str, Any] = {"data": data}
+        else:
+            kwargs = {"source": data}
+
+        kwargs.update(
+            {
+                "validators": validators,
+                "schema": schema,
+                "auto_schema": auto_schema,
+                "parallel": parallel,
+            }
+        )
 
-        # Add per-validator parameters if provided
-        if validator_params:
-            kwargs["validator_params"] = validator_params
+        # Add per-validator configuration if provided (truthound 2.x uses validator_config)
+        if validator_config:
+            kwargs["validator_config"] = validator_config
 
         # Only add optional params if explicitly set
-        if columns is not None:
-            kwargs["columns"] = columns
         if min_severity is not None:
             kwargs["min_severity"] = min_severity
-        if strict:
-            kwargs["strict"] = strict
         if max_workers is not None:
             kwargs["max_workers"] = max_workers
         if pushdown is not None:
             kwargs["pushdown"] = pushdown
 
-        func = partial(th.check, data, **kwargs)
+        func = partial(th.check, **kwargs)
 
         loop = asyncio.get_event_loop()
         result = await loop.run_in_executor(self._executor, func)
@@ -369,7 +617,7 @@ class TruthoundAdapter:
 
     async def learn(
         self,
-        source: str,
+        source: DataInput,
         *,
         infer_constraints: bool = True,
         categorical_threshold: int | None = None,
@@ -378,23 +626,36 @@ class TruthoundAdapter:
         """Learn schema from data asynchronously.
 
         Uses truthound's th.learn() to analyze data and generate schema.
-        Supports all th.learn() parameters for maximum flexibility.
+        If sample_size is provided, delegates to learn_with_sampling() which
+        handles dashboard-level sampling before calling th.learn().
+
+        Note: th.learn() only supports (data, infer_constraints, categorical_threshold).
+        sample_size is handled at dashboard level, not passed to truthound.
 
         Args:
-            source: Data source path.
+            source: Data source - can be:
+                - File path string
+                - DataSource object
             infer_constraints: If True, infers constraints (min/max, allowed values)
                 from data statistics.
             categorical_threshold: Maximum unique values for categorical detection.
                 Columns with unique values <= threshold are treated as categorical
                 and will have allowed_values inferred. If None, uses truthound
                 default (20).
-            sample_size: Number of rows to sample for large datasets.
-                If None, uses all rows. Sampling improves performance but may
-                miss rare values.
+            sample_size: Sample size for large datasets. Handled at dashboard level
+                by pre-sampling data before passing to th.learn().
 
         Returns:
             LearnResult with schema information.
         """
+        if sample_size is not None:
+            return await self.learn_with_sampling(
+                source,
+                infer_constraints=infer_constraints,
+                categorical_threshold=categorical_threshold,
+                sample_size=sample_size,
+            )
+
         import truthound as th
 
         # Build kwargs dynamically to let truthound use its defaults when not specified
@@ -402,8 +663,6 @@ class TruthoundAdapter:
 
         if categorical_threshold is not None:
             kwargs["categorical_threshold"] = categorical_threshold
-        if sample_size is not None:
-            kwargs["sample_size"] = sample_size
 
         func = partial(th.learn, source, **kwargs)
 
@@ -414,75 +673,329 @@ class TruthoundAdapter:
 
     async def profile(
         self,
-        source: str,
-        *,
-        sample_size: int | None = None,
+        source: DataInput,
     ) -> ProfileResult:
         """Run data profiling asynchronously.
 
+        Note: truthound's th.profile() only accepts (data, source) parameters.
+        Advanced configuration options are NOT supported by the underlying library.
+
         Args:
-            source: Data source path.
-            sample_size: Maximum number of rows to sample for profiling.
-                If None, profiles all data. Useful for large datasets.
+            source: Data source - can be:
+                - File path string
+                - DataSource object
 
         Returns:
             ProfileResult with profiling information.
         """
         import truthound as th
 
-        # Build kwargs dynamically to let truthound use its defaults
-        kwargs: dict[str, Any] = {}
-        if sample_size is not None:
-            kwargs["sample_size"] = sample_size
-
-        func = partial(th.profile, source, **kwargs)
+        func = partial(th.profile, source)
 
         loop = asyncio.get_event_loop()
         result = await loop.run_in_executor(self._executor, func)
+        return self._convert_profile_result(result)
+
+    async def profile_advanced(
+        self,
+        source: DataInput,
+        *,
+        config: dict[str, Any] | None = None,
+    ) -> ProfileResult:
+        """Run advanced data profiling with full ProfilerConfig support.
+
+        This method provides direct access to all ProfilerConfig options
+        through a configuration dictionary.
+
+        Note: DataProfiler.profile() only accepts LazyFrame, so file paths
+        are converted to LazyFrame first. For simple profiling without
+        advanced config, use profile() method instead.
+
+        Args:
+            source: Data source - file path string or DataSource object.
+            config: ProfilerConfig options as dictionary. Supported keys:
+                - sample_size: int | None (max rows to sample)
+                - random_seed: int (default 42)
+                - include_patterns: bool (default True)
+                - include_correlations: bool (default False)
+                - include_distributions: bool (default True)
+                - top_n_values: int (default 10)
+                - pattern_sample_size: int (default 1000)
+                - correlation_threshold: float (default 0.7)
+                - min_pattern_match_ratio: float (default 0.8)
+                - n_jobs: int (default 1)
+
+        Returns:
+            ProfileResult with comprehensive profiling information.
+
+        Raises:
+            ImportError: If truthound.profiler module is not available.
+        """
+        import polars as pl
+
+        from truthound.profiler import DataProfiler, ProfilerConfig
+
+        config = config or {}
+
+        profiler_config = ProfilerConfig(
+            sample_size=config.get("sample_size"),
+            random_seed=config.get("random_seed", 42),
+            include_patterns=config.get("include_patterns", True),
+            include_correlations=config.get("include_correlations", False),
+            include_distributions=config.get("include_distributions", True),
+            top_n_values=config.get("top_n_values", 10),
+            pattern_sample_size=config.get("pattern_sample_size", 1000),
+            correlation_threshold=config.get("correlation_threshold", 0.7),
+            min_pattern_match_ratio=config.get("min_pattern_match_ratio", 0.8),
+            n_jobs=config.get("n_jobs", 1),
+        )
+
+        profiler = DataProfiler(config=profiler_config)
+
+        # DataProfiler.profile() only accepts LazyFrame
+        # Convert file path to LazyFrame
+        if isinstance(source, str):
+            # Determine file format and create LazyFrame
+            source_lower = source.lower()
+            if source_lower.endswith(".csv"):
+                lf = pl.scan_csv(source)
+            elif source_lower.endswith(".parquet"):
+                lf = pl.scan_parquet(source)
+            elif source_lower.endswith((".json", ".ndjson", ".jsonl")):
+                lf = pl.scan_ndjson(source)
+            else:
+                # Fallback to th.profile() for unsupported formats
+                import truthound as th
+
+                func = partial(th.profile, source)
+                loop = asyncio.get_event_loop()
+                result = await loop.run_in_executor(self._executor, func)
+                return self._convert_profile_result(result)
+
+            func = partial(profiler.profile, lf, name=source, source=source)
+        elif hasattr(source, "lazy"):
+            # DataFrame with .lazy() method
+            func = partial(profiler.profile, source.lazy())
+        elif hasattr(source, "collect"):
+            # Already a LazyFrame
+            func = partial(profiler.profile, source)
+        else:
+            # Fallback to th.profile() for other types
+            import truthound as th
 
+            func = partial(th.profile, source)
+            loop = asyncio.get_event_loop()
+            result = await loop.run_in_executor(self._executor, func)
+            return self._convert_profile_result(result)
+
+        loop = asyncio.get_event_loop()
+        result = await loop.run_in_executor(self._executor, func)
         return self._convert_profile_result(result)
 
-    async def scan(
+    async def generate_suite(
         self,
-        data: str,
+        profile: ProfileResult | dict[str, Any],
         *,
-        columns: list[str] | None = None,
-        regulations: list[str] | None = None,
-        min_confidence: float = 0.8,
-    ) -> ScanResult:
-        """Run PII scan on data asynchronously.
+        strictness: str = "medium",
+        preset: str = "default",
+        include: list[str] | None = None,
+        exclude: list[str] | None = None,
+        output_format: str = "yaml",
+    ) -> GenerateSuiteResult:
+        """Generate validation suite from profile.
 
-        Uses truthound's th.scan() to detect personally identifiable information
-        and check compliance with privacy regulations.
+        Uses truthound's generate_suite() to automatically create validation
+        rules based on profiled data characteristics.
 
         Args:
-            data: Data source path (CSV, Parquet, etc.).
-            columns: Optional list of columns to scan. If None, scans all columns.
-            regulations: Optional list of regulations to check compliance.
-                Supported: "gdpr", "ccpa", "lgpd"
-            min_confidence: Minimum confidence threshold for PII detection (0.0-1.0).
-                Default is 0.8.
+            profile: Profile result from profile() or profile_advanced(),
+                or a dictionary representation of a profile.
+            strictness: Strictness level for rule generation:
+                - "loose": Permissive thresholds, fewer rules
+                - "medium": Balanced defaults (default)
+                - "strict": Tight thresholds, comprehensive rules
+            preset: Configuration preset for rule generation:
+                - "default": General purpose
+                - "strict": Production data
+                - "loose": Development/testing
+                - "minimal": Essential rules only
+                - "comprehensive": All available rules
+                - "ci_cd": Optimized for CI/CD pipelines
+                - "schema_only": Structure validation only
+                - "format_only": Format/pattern rules only
+            include: List of rule categories to include (None = all).
+                Categories: schema, stats, pattern, completeness, uniqueness, distribution
+            exclude: List of rule categories to exclude.
+            output_format: Output format ("yaml", "json", "python").
 
         Returns:
-            ScanResult with PII findings and regulation violations.
+            GenerateSuiteResult with generated rules.
 
         Raises:
-            ImportError: If truthound is not installed.
-            FileNotFoundError: If data file doesn't exist.
+            ImportError: If truthound.profiler module is not available.
         """
-        import truthound as th
+        from truthound.profiler import generate_suite
+        from truthound.profiler.generators import Strictness
+
+        # Convert strictness string to enum
+        strictness_map = {
+            "loose": Strictness.LOOSE,
+            "medium": Strictness.MEDIUM,
+            "strict": Strictness.STRICT,
+        }
+        strictness_enum = strictness_map.get(strictness.lower(), Strictness.MEDIUM)
+
+        # Convert ProfileResult to dict if needed
+        if isinstance(profile, ProfileResult):
+            profile_data = profile.to_dict()
+        else:
+            profile_data = profile
 
-        # Build kwargs dynamically to let truthound use its defaults
+        # Build kwargs
         kwargs: dict[str, Any] = {
-            "min_confidence": min_confidence,
+            "strictness": strictness_enum,
+            "preset": preset,
         }
+        if include:
+            kwargs["include"] = include
+        if exclude:
+            kwargs["exclude"] = exclude
 
-        if columns is not None:
-            kwargs["columns"] = columns
-        if regulations is not None:
-            kwargs["regulations"] = regulations
+        # Generate suite in thread pool
+        def _generate():
+            return generate_suite(profile_data, **kwargs)
+
+        loop = asyncio.get_event_loop()
+        suite = await loop.run_in_executor(self._executor, _generate)
+
+        return self._convert_suite_result(suite, strictness, output_format)
+
+    async def generate_suite_from_source(
+        self,
+        source: DataInput,
+        *,
+        strictness: str = "medium",
+        preset: str = "default",
+        include: list[str] | None = None,
+        exclude: list[str] | None = None,
+        sample_size: int | None = None,
+        include_patterns: bool = True,
+    ) -> GenerateSuiteResult:
+        """Profile a source and generate validation suite in one step.
+
+        Convenience method that combines profile() and generate_suite().
+
+        Args:
+            source: Data source - file path string or DataSource object.
+            strictness: Strictness level ("loose", "medium", "strict").
+            preset: Rule generation preset.
+            include: Rule categories to include.
+            exclude: Rule categories to exclude.
+            sample_size: Number of rows to sample for profiling.
+            include_patterns: Enable pattern detection during profiling.
+
+        Returns:
+            GenerateSuiteResult with generated rules.
+        """
+        # Profile the source first
+        profile = await self.profile(
+            source,
+            sample_size=sample_size,
+            include_patterns=include_patterns,
+        )
+
+        # Generate suite from profile
+        return await self.generate_suite(
+            profile,
+            strictness=strictness,
+            preset=preset,
+            include=include,
+            exclude=exclude,
+        )
+
+    def _convert_suite_result(
+        self,
+        suite: Any,
+        strictness: str,
+        output_format: str,
+    ) -> GenerateSuiteResult:
+        """Convert truthound ValidationSuite to GenerateSuiteResult.
+
+        Args:
+            suite: ValidationSuite from generate_suite().
+            strictness: Strictness level used.
+            output_format: Requested output format.
+
+        Returns:
+            GenerateSuiteResult.
+        """
+        # Extract rules from suite
+        rules = []
+        categories = set()
+
+        if hasattr(suite, "rules"):
+            for rule in suite.rules:
+                rule_dict = {
+                    "name": getattr(rule, "name", ""),
+                    "validator": getattr(rule, "validator", ""),
+                    "column": getattr(rule, "column", None),
+                    "params": getattr(rule, "params", {}),
+                    "severity": getattr(rule, "severity", "medium"),
+                    "category": getattr(rule, "category", "unknown"),
+                }
+                rules.append(rule_dict)
+                if rule_dict["category"]:
+                    categories.add(rule_dict["category"])
+
+        # Generate YAML content
+        yaml_content = ""
+        if hasattr(suite, "to_yaml"):
+            yaml_content = suite.to_yaml()
+        else:
+            yaml_content = yaml.dump(
+                {"rules": rules},
+                default_flow_style=False,
+                sort_keys=False,
+                allow_unicode=True,
+            )
+
+        # Generate JSON content
+        json_content = {"rules": rules}
+        if hasattr(suite, "to_dict"):
+            json_content = suite.to_dict()
+
+        return GenerateSuiteResult(
+            rules=rules,
+            rule_count=len(rules),
+            categories=sorted(categories),
+            strictness=strictness,
+            yaml_content=yaml_content,
+            json_content=json_content,
+        )
+
+    async def scan(self, data: DataInput) -> ScanResult:
+        """Run PII scan on data asynchronously.
+
+        Uses truthound's th.scan() to detect personally identifiable information.
+
+        Note: truthound's th.scan() does not support any configuration parameters.
+        The scan runs on all columns with default settings.
+
+        Args:
+            data: Data source - can be:
+                - File path string (CSV, Parquet, etc.)
+                - DataSource object
+
+        Returns:
+            ScanResult with PII findings.
+
+        Raises:
+            ImportError: If truthound is not installed.
+            FileNotFoundError: If data file doesn't exist.
+        """
+        import truthound as th
 
-        func = partial(th.scan, data, **kwargs)
+        func = partial(th.scan, data)
 
         loop = asyncio.get_event_loop()
         result = await loop.run_in_executor(self._executor, func)
@@ -491,20 +1004,19 @@ class TruthoundAdapter:
 
     async def compare(
         self,
-        baseline: str,
-        current: str,
+        baseline: DataInput,
+        current: DataInput,
         *,
         columns: list[str] | None = None,
         method: str = "auto",
         threshold: float | None = None,
-        correction: str | None = None,
         sample_size: int | None = None,
     ) -> CompareResult:
         """Compare two datasets for drift detection.
 
         Args:
-            baseline: Reference data path.
-            current: Current data path to compare.
+            baseline: Reference data - can be path string or DataSource.
+            current: Current data to compare - can be path string or DataSource.
             columns: Optional list of columns to compare. If None, all common columns.
             method: Detection method. Supported methods:
                 - "auto": Smart selection (numeric → PSI, categorical → chi2)
@@ -518,12 +1030,6 @@ class TruthoundAdapter:
                 - "anderson": Anderson-Darling (tail-weighted)
             threshold: Optional custom threshold for drift detection.
                 Defaults vary by method: KS/chi2/cvm/anderson=0.05, PSI/JS/KL/wasserstein=0.1
-            correction: Multiple testing correction method:
-                - None: Use truthound default (bh for multiple columns)
-                - "none": No correction
-                - "bonferroni": Conservative, independent tests
-                - "holm": Sequential adjustment
-                - "bh": Benjamini-Hochberg FDR control
             sample_size: Optional sample size for large datasets.
 
         Returns:
@@ -531,17 +1037,13 @@ class TruthoundAdapter:
         """
         import truthound as th
 
-        # Build kwargs dynamically to avoid passing None for optional params
         kwargs: dict[str, Any] = {
             "columns": columns,
             "method": method,
         }
 
-        # Only add optional params if explicitly set
         if threshold is not None:
             kwargs["threshold"] = threshold
-        if correction is not None:
-            kwargs["correction"] = correction
         if sample_size is not None:
             kwargs["sample_size"] = sample_size
 
@@ -554,7 +1056,7 @@ class TruthoundAdapter:
 
     async def mask(
         self,
-        data: str,
+        data: DataInput,
         output: str,
         *,
         columns: list[str] | None = None,
@@ -566,7 +1068,9 @@ class TruthoundAdapter:
         three strategies: redact, hash, and fake.
 
         Args:
-            data: Data source path (CSV, Parquet, etc.).
+            data: Data source - can be:
+                - File path string (CSV, Parquet, etc.)
+                - DataSource object
             output: Output file path for the masked data.
             columns: Optional list of columns to mask. If None, auto-detects PII.
             strategy: Masking strategy:
@@ -607,10 +1111,10 @@ class TruthoundAdapter:
 
     async def check_with_sampling(
         self,
-        data: str,
+        data: DataInput,
         *,
         validators: list[str] | None = None,
-        validator_params: dict[str, dict[str, Any]] | None = None,
+        validator_config: dict[str, dict[str, Any]] | None = None,
         schema: str | None = None,
         auto_schema: bool = False,
         columns: list[str] | None = None,
@@ -628,9 +1132,14 @@ class TruthoundAdapter:
         before running validation, which significantly improves performance
         while maintaining validation accuracy for most use cases.
 
+        Note: Sampling is only applied to file-based sources. DataSource
+        objects handle their own data fetching and should use query-level
+        sampling if needed.
+
         Args:
-            data: Data source path (CSV, Parquet, etc.).
+            data: Data source - can be file path or DataSource.
             validators: Optional list of validator names to run.
+            validator_config: Optional dict of per-validator configuration.
             schema: Optional path to schema YAML file.
             auto_schema: If True, auto-learns schema for validation.
             columns: Columns to validate. If None, validates all columns.
@@ -649,40 +1158,42 @@ class TruthoundAdapter:
             The result.row_count reflects the sampled row count when sampling
             was performed. Check the sampling metadata for original row count.
         """
-        from truthound_dashboard.core.sampling import SamplingMethod, get_sampler
-
-        sampler = get_sampler()
-
-        # Check if sampling is needed and perform if so
-        path = Path(data)
-        if path.exists() and sampler.needs_sampling(path):
-            # Determine sampling method
-            method = None
-            if sampling_method:
-                try:
-                    method = SamplingMethod(sampling_method)
-                except ValueError:
-                    logger.warning(f"Unknown sampling method: {sampling_method}")
-
-            # Perform sampling
-            sample_result = await sampler.auto_sample(
-                path,
-                n=sample_size,
-                method=method,
-            )
-
-            if sample_result.was_sampled:
-                logger.info(
-                    f"Sampled {sample_result.sampled_rows} rows from "
-                    f"{sample_result.original_rows} ({sample_result.size_reduction_pct:.1f}% reduction)"
+        # Only apply sampling to file paths
+        if isinstance(data, str):
+            from truthound_dashboard.core.sampling import SamplingMethod, get_sampler
+
+            sampler = get_sampler()
+
+            # Check if sampling is needed and perform if so
+            path = Path(data)
+            if path.exists() and sampler.needs_sampling(path):
+                # Determine sampling method
+                method = None
+                if sampling_method:
+                    try:
+                        method = SamplingMethod(sampling_method)
+                    except ValueError:
+                        logger.warning(f"Unknown sampling method: {sampling_method}")
+
+                # Perform sampling
+                sample_result = await sampler.auto_sample(
+                    path,
+                    n=sample_size,
+                    method=method,
                 )
-                data = sample_result.sampled_path
+
+                if sample_result.was_sampled:
+                    logger.info(
+                        f"Sampled {sample_result.sampled_rows} rows from "
+                        f"{sample_result.original_rows} ({sample_result.size_reduction_pct:.1f}% reduction)"
+                    )
+                    data = sample_result.sampled_path
 
         # Run validation on (possibly sampled) data
         return await self.check(
             data,
             validators=validators,
-            validator_params=validator_params,
+            validator_config=validator_config,
             schema=schema,
             auto_schema=auto_schema,
             columns=columns,
@@ -695,7 +1206,7 @@ class TruthoundAdapter:
 
     async def learn_with_sampling(
         self,
-        source: str,
+        source: DataInput,
         *,
         infer_constraints: bool = True,
         categorical_threshold: int | None = None,
@@ -706,8 +1217,10 @@ class TruthoundAdapter:
         This method first applies dashboard-level sampling for very large files,
         then passes the sample_size to th.learn() if specified.
 
+        Note: Sampling is only applied to file-based sources.
+
         Args:
-            source: Data source path.
+            source: Data source - can be file path or DataSource.
             infer_constraints: If True, infer constraints from statistics.
             categorical_threshold: Maximum unique values for categorical detection.
             sample_size: Number of rows to sample. Used both for dashboard sampling
@@ -716,57 +1229,138 @@ class TruthoundAdapter:
         Returns:
             LearnResult with schema information.
         """
-        from truthound_dashboard.core.sampling import get_sampler
-
-        sampler = get_sampler()
-
-        # Sample if needed (dashboard-level sampling for very large files)
-        path = Path(source)
-        if path.exists() and sampler.needs_sampling(path):
-            sample_result = await sampler.auto_sample(path, n=sample_size)
-            if sample_result.was_sampled:
-                logger.info(
-                    f"Sampled {sample_result.sampled_rows} rows for schema learning"
-                )
-                source = sample_result.sampled_path
-
+        # Only apply sampling to file paths
+        if isinstance(source, str):
+            from truthound_dashboard.core.sampling import get_sampler
+
+            sampler = get_sampler()
+
+            path = Path(source)
+            if path.exists() and sampler.needs_sampling(path):
+                sample_result = await sampler.auto_sample(path, n=sample_size)
+                if sample_result.was_sampled:
+                    logger.info(
+                        f"Sampled {sample_result.sampled_rows} rows for schema learning"
+                    )
+                    source = sample_result.sampled_path
+
+        # sample_size already handled by dashboard-level sampling above,
+        # do NOT pass it to self.learn() — th.learn() doesn't support it
         return await self.learn(
             source,
             infer_constraints=infer_constraints,
             categorical_threshold=categorical_threshold,
-            sample_size=sample_size,
         )
 
     async def profile_with_sampling(
         self,
-        source: str,
+        source: DataInput,
         *,
         sample_size: int | None = None,
+        include_patterns: bool = True,
+        include_correlations: bool = False,
     ) -> ProfileResult:
         """Run data profiling with automatic sampling for large datasets.
 
+        Note: Sampling is only applied to file-based sources.
+
         Args:
-            source: Data source path.
+            source: Data source - can be file path or DataSource.
             sample_size: Number of rows to sample. Uses config default if not specified.
+            include_patterns: Enable pattern detection. Default True.
+            include_correlations: Calculate correlations. Default False.
 
         Returns:
             ProfileResult with profiling information.
         """
-        from truthound_dashboard.core.sampling import get_sampler
+        # Only apply sampling to file paths
+        if isinstance(source, str):
+            from truthound_dashboard.core.sampling import get_sampler
+
+            sampler = get_sampler()
+
+            path = Path(source)
+            if path.exists() and sampler.needs_sampling(path):
+                sample_result = await sampler.auto_sample(path, n=sample_size)
+                if sample_result.was_sampled:
+                    logger.info(
+                        f"Sampled {sample_result.sampled_rows} rows for profiling"
+                    )
+                    source = sample_result.sampled_path
+
+        return await self.profile(
+            source,
+            sample_size=sample_size,
+            include_patterns=include_patterns,
+            include_correlations=include_correlations,
+        )
+
+    async def check_from_config(
+        self,
+        source_config: "SourceConfig | dict[str, Any]",
+        *,
+        validators: list[str] | None = None,
+        validator_config: dict[str, dict[str, Any]] | None = None,
+        schema: str | None = None,
+        auto_schema: bool = False,
+        columns: list[str] | None = None,
+        min_severity: str | None = None,
+        strict: bool = False,
+        parallel: bool = False,
+        max_workers: int | None = None,
+        pushdown: bool | None = None,
+    ) -> CheckResult:
+        """Run validation using source configuration.
 
-        sampler = get_sampler()
+        This convenience method creates a DataSource from config
+        and runs validation.
 
-        # Sample if needed
-        path = Path(source)
-        if path.exists() and sampler.needs_sampling(path):
-            sample_result = await sampler.auto_sample(path, n=sample_size)
-            if sample_result.was_sampled:
-                logger.info(
-                    f"Sampled {sample_result.sampled_rows} rows for profiling"
-                )
-                source = sample_result.sampled_path
+        Args:
+            source_config: Source configuration (SourceConfig or dict).
+            validators: Optional list of validator names to run.
+            validator_config: Optional dict of per-validator configuration.
+            schema: Optional path to schema YAML file.
+            auto_schema: If True, auto-learns schema for validation.
+            columns: Columns to validate.
+            min_severity: Minimum severity to report.
+            strict: If True, raises exception on validation failures.
+            parallel: If True, uses parallel execution.
+            max_workers: Max threads for parallel execution.
+            pushdown: Enable query pushdown for SQL sources.
+
+        Returns:
+            CheckResult with validation results.
+        """
+        from truthound_dashboard.core.datasource_factory import (
+            SourceConfig,
+            SourceType,
+            create_datasource,
+        )
+
+        if isinstance(source_config, dict):
+            config = SourceConfig.from_dict(source_config)
+        else:
+            config = source_config
+
+        # For file sources, use path directly
+        if SourceType.is_file_type(config.source_type) and config.path:
+            data: DataInput = config.path
+        else:
+            data = create_datasource(config)
 
-        return await self.profile(source)
+        return await self.check(
+            data,
+            validators=validators,
+            validator_config=validator_config,
+            schema=schema,
+            auto_schema=auto_schema,
+            columns=columns,
+            min_severity=min_severity,
+            strict=strict,
+            parallel=parallel,
+            max_workers=max_workers,
+            pushdown=pushdown,
+        )
 
     def _convert_check_result(self, result: Any) -> CheckResult:
         """Convert truthound Report to CheckResult.
@@ -779,7 +1373,15 @@ class TruthoundAdapter:
         - has_issues: bool
         - has_critical: bool
         - has_high: bool
+
+        Also handles truthound 2.x ValidationResult format with:
+        - run_id: str
+        - run_time: datetime
+        - results: list[ValidatorResult]
+        - statistics: ResultStatistics
         """
+        from datetime import datetime
+
         issues = result.issues
         severity_counts = {"critical": 0, "high": 0, "medium": 0, "low": 0}
 
@@ -797,10 +1399,19 @@ class TruthoundAdapter:
                 "details": getattr(issue, "details", None),
                 "expected": getattr(issue, "expected", None),
                 "actual": getattr(issue, "actual", None),
+                "validator_name": getattr(issue, "validator_name", issue.issue_type),
+                "message": getattr(issue, "message", ""),
+                "sample_values": getattr(issue, "sample_values", None),
             }
             for issue in issues
         ]
 
+        # Extract run_id and run_time if available (truthound 2.x)
+        run_id = getattr(result, "run_id", None)
+        run_time = getattr(result, "run_time", None)
+        if run_time is None:
+            run_time = datetime.now()
+
         return CheckResult(
             passed=not result.has_issues,
             has_critical=result.has_critical,
@@ -814,6 +1425,9 @@ class TruthoundAdapter:
             row_count=result.row_count,
             column_count=result.column_count,
             issues=converted_issues,
+            run_id=run_id,
+            run_time=run_time,
+            _raw_result=result,  # Store raw result for reporter integration
         )
 
     def _convert_learn_result(self, result: Any) -> LearnResult:
@@ -842,35 +1456,241 @@ class TruthoundAdapter:
         )
 
     def _convert_profile_result(self, result: Any) -> ProfileResult:
-        """Convert truthound ProfileReport to ProfileResult.
+        """Convert truthound TableProfile/ProfileReport to ProfileResult.
+
+        The truthound TableProfile (new API) contains:
+        - name: str
+        - row_count: int
+        - column_count: int
+        - estimated_memory_bytes: int
+        - columns: tuple[ColumnProfile, ...]
+        - duplicate_row_count: int
+        - duplicate_row_ratio: float
+        - correlations: tuple[tuple[str, str, float], ...]
+        - source: str
+        - profiled_at: datetime
+        - profile_duration_ms: float
+
+        Each ColumnProfile contains:
+        - name: str
+        - physical_type: str
+        - inferred_type: DataType enum
+        - row_count, null_count, null_ratio, empty_string_count
+        - distinct_count, unique_ratio, is_unique, is_constant
+        - distribution: DistributionStats | None
+        - top_values, bottom_values: tuple[ValueFrequency, ...]
+        - min_length, max_length, avg_length (string columns)
+        - detected_patterns: tuple[PatternMatch, ...]
+        - min_date, max_date, date_gaps (datetime columns)
+        - suggested_validators: tuple[str, ...]
+        - profile_duration_ms: float
+
+        Also supports legacy ProfileReport format for backward compatibility.
+        """
+        # Check if this is the new TableProfile or legacy ProfileReport
+        if hasattr(result, "estimated_memory_bytes"):
+            # New TableProfile format
+            return self._convert_table_profile(result)
+        else:
+            # Legacy ProfileReport format - convert to new structure
+            return self._convert_legacy_profile(result)
+
+    def _convert_table_profile(self, result: Any) -> ProfileResult:
+        """Convert new truthound TableProfile to ProfileResult."""
+        from datetime import datetime
+
+        columns = []
+        for col in result.columns:
+            # Extract distribution stats if present
+            distribution = None
+            if col.distribution:
+                distribution = {
+                    "mean": getattr(col.distribution, "mean", None),
+                    "std": getattr(col.distribution, "std", None),
+                    "min": getattr(col.distribution, "min", None),
+                    "max": getattr(col.distribution, "max", None),
+                    "median": getattr(col.distribution, "median", None),
+                    "q1": getattr(col.distribution, "q1", None),
+                    "q3": getattr(col.distribution, "q3", None),
+                    "skewness": getattr(col.distribution, "skewness", None),
+                    "kurtosis": getattr(col.distribution, "kurtosis", None),
+                }
+
+            # Convert top_values
+            top_values = None
+            if col.top_values:
+                top_values = [
+                    {
+                        "value": str(v.value) if v.value is not None else None,
+                        "count": v.count,
+                        "ratio": v.ratio,
+                    }
+                    for v in col.top_values
+                ]
+
+            # Convert bottom_values
+            bottom_values = None
+            if col.bottom_values:
+                bottom_values = [
+                    {
+                        "value": str(v.value) if v.value is not None else None,
+                        "count": v.count,
+                        "ratio": v.ratio,
+                    }
+                    for v in col.bottom_values
+                ]
+
+            # Convert detected_patterns
+            detected_patterns = None
+            if col.detected_patterns:
+                detected_patterns = [
+                    {
+                        "pattern": getattr(p, "pattern", None),
+                        "regex": getattr(p, "regex", None),
+                        "match_ratio": getattr(p, "match_ratio", 0.0),
+                        "sample_matches": list(getattr(p, "sample_matches", [])),
+                    }
+                    for p in col.detected_patterns
+                ]
+
+            # Get inferred type value
+            inferred_type = "unknown"
+            if hasattr(col, "inferred_type"):
+                inferred_type = (
+                    col.inferred_type.value
+                    if hasattr(col.inferred_type, "value")
+                    else str(col.inferred_type)
+                )
+
+            # Convert datetime fields to ISO strings
+            min_date = None
+            max_date = None
+            if col.min_date:
+                min_date = (
+                    col.min_date.isoformat()
+                    if isinstance(col.min_date, datetime)
+                    else str(col.min_date)
+                )
+            if col.max_date:
+                max_date = (
+                    col.max_date.isoformat()
+                    if isinstance(col.max_date, datetime)
+                    else str(col.max_date)
+                )
+
+            col_result = ColumnProfileResult(
+                name=col.name,
+                physical_type=col.physical_type,
+                inferred_type=inferred_type,
+                row_count=col.row_count,
+                null_count=col.null_count,
+                null_ratio=col.null_ratio,
+                empty_string_count=col.empty_string_count,
+                distinct_count=col.distinct_count,
+                unique_ratio=col.unique_ratio,
+                is_unique=col.is_unique,
+                is_constant=col.is_constant,
+                distribution=distribution,
+                top_values=top_values,
+                bottom_values=bottom_values,
+                min_length=col.min_length,
+                max_length=col.max_length,
+                avg_length=col.avg_length,
+                detected_patterns=detected_patterns,
+                min_date=min_date,
+                max_date=max_date,
+                date_gaps=col.date_gaps,
+                suggested_validators=list(col.suggested_validators)
+                if col.suggested_validators
+                else None,
+                profile_duration_ms=col.profile_duration_ms,
+            )
+            columns.append(col_result)
+
+        # Convert correlations
+        correlations = None
+        if result.correlations:
+            correlations = [
+                (c[0], c[1], c[2]) for c in result.correlations
+            ]
+
+        # Get profiled_at as ISO string
+        profiled_at = None
+        if hasattr(result, "profiled_at") and result.profiled_at:
+            profiled_at = (
+                result.profiled_at.isoformat()
+                if isinstance(result.profiled_at, datetime)
+                else str(result.profiled_at)
+            )
+
+        return ProfileResult(
+            name=getattr(result, "name", ""),
+            source=getattr(result, "source", ""),
+            row_count=result.row_count,
+            column_count=result.column_count,
+            estimated_memory_bytes=result.estimated_memory_bytes,
+            columns=columns,
+            duplicate_row_count=result.duplicate_row_count,
+            duplicate_row_ratio=result.duplicate_row_ratio,
+            correlations=correlations,
+            profiled_at=profiled_at,
+            profile_duration_ms=getattr(result, "profile_duration_ms", 0.0),
+            size_bytes=result.estimated_memory_bytes,
+        )
+
+    def _convert_legacy_profile(self, result: Any) -> ProfileResult:
+        """Convert legacy truthound ProfileReport to ProfileResult.
 
-        The truthound ProfileReport contains:
+        Legacy ProfileReport contains:
         - source: str
         - row_count: int
         - column_count: int
         - size_bytes: int
-        - columns: list[dict]
+        - columns: list[dict] with name, dtype, null_pct, unique_pct, min, max, mean, std
         """
-        columns = [
-            {
-                "name": col["name"],
-                "dtype": col["dtype"],
-                "null_pct": col.get("null_pct", "0%"),
-                "unique_pct": col.get("unique_pct", "0%"),
-                "min": col.get("min"),
-                "max": col.get("max"),
-                "mean": col.get("mean"),
-                "std": col.get("std"),
-            }
-            for col in result.columns
-        ]
+        columns = []
+        for col in result.columns:
+            # Parse null_pct and unique_pct
+            null_ratio = 0.0
+            unique_ratio = 0.0
+            if isinstance(col.get("null_pct"), str):
+                null_ratio = float(col["null_pct"].rstrip("%")) / 100.0
+            elif isinstance(col.get("null_pct"), (int, float)):
+                null_ratio = float(col["null_pct"])
+            if isinstance(col.get("unique_pct"), str):
+                unique_ratio = float(col["unique_pct"].rstrip("%")) / 100.0
+            elif isinstance(col.get("unique_pct"), (int, float)):
+                unique_ratio = float(col["unique_pct"])
+
+            # Build distribution if numeric stats present
+            distribution = None
+            if col.get("min") is not None or col.get("mean") is not None:
+                distribution = {
+                    "min": col.get("min"),
+                    "max": col.get("max"),
+                    "mean": col.get("mean"),
+                    "std": col.get("std"),
+                }
+
+            col_result = ColumnProfileResult(
+                name=col["name"],
+                physical_type=col.get("dtype", "unknown"),
+                inferred_type=col.get("dtype", "unknown"),
+                row_count=result.row_count,
+                null_ratio=null_ratio,
+                unique_ratio=unique_ratio,
+                distribution=distribution,
+            )
+            columns.append(col_result)
 
         return ProfileResult(
+            name=getattr(result, "source", ""),
             source=result.source,
             row_count=result.row_count,
             column_count=result.column_count,
-            size_bytes=result.size_bytes,
+            estimated_memory_bytes=getattr(result, "size_bytes", 0),
             columns=columns,
+            size_bytes=getattr(result, "size_bytes", 0),
         )
 
     def _convert_scan_result(self, result: Any) -> ScanResult:
@@ -887,7 +1707,7 @@ class TruthoundAdapter:
         Each PIIFinding has:
         - column: str
         - pii_type: str
-        - confidence: float
+        - confidence: float (0-100)
         - sample_count: int
         - sample_values: list[str] (optional)
 
@@ -897,19 +1717,41 @@ class TruthoundAdapter:
         - pii_type: str
         - message: str
         - severity: str (optional)
+
+        Args:
+            result: truthound PIIReport object.
+
+        Returns:
+            ScanResult with PII findings.
         """
         # Convert findings to dictionaries
         findings = []
         columns_with_pii = set()
         for finding in result.findings:
-            columns_with_pii.add(finding.column)
+            # Handle both dict and object-style findings
+            if isinstance(finding, dict):
+                confidence = finding.get("confidence", 0)
+                column = finding.get("column", "")
+                pii_type = finding.get("pii_type", "unknown")
+                sample_count = finding.get("count", finding.get("sample_count", 0))
+                sample_values = finding.get("sample_values")
+            else:
+                confidence = getattr(finding, "confidence", 0)
+                column = getattr(finding, "column", "")
+                pii_type = getattr(finding, "pii_type", "unknown")
+                sample_count = getattr(finding, "sample_count", getattr(finding, "count", 0))
+                sample_values = getattr(finding, "sample_values", None)
+
+            columns_with_pii.add(column)
+            # Normalize confidence to 0-1 range if it's in 0-100 range
+            normalized_confidence = confidence / 100.0 if confidence > 1 else confidence
             findings.append(
                 {
-                    "column": finding.column,
-                    "pii_type": finding.pii_type,
-                    "confidence": finding.confidence,
-                    "sample_count": finding.sample_count,
-                    "sample_values": getattr(finding, "sample_values", None),
+                    "column": column,
+                    "pii_type": pii_type,
+                    "confidence": normalized_confidence,
+                    "sample_count": sample_count,
+                    "sample_values": sample_values,
                 }
             )
 
@@ -926,11 +1768,14 @@ class TruthoundAdapter:
                 }
             )
 
+        # Get column_count with fallback (not present in some truthound versions)
+        column_count = getattr(result, "column_count", len(columns_with_pii) if columns_with_pii else 0)
+
         return ScanResult(
             source=result.source,
             row_count=result.row_count,
-            column_count=result.column_count,
-            total_columns_scanned=result.column_count,
+            column_count=column_count,
+            total_columns_scanned=column_count,
             columns_with_pii=len(columns_with_pii),
             total_findings=len(findings),
             has_violations=getattr(result, "has_violations", len(violations) > 0),
@@ -992,7 +1837,7 @@ class TruthoundAdapter:
 
     def _convert_mask_result(
         self,
-        source: str,
+        source: DataInput,
         output: str,
         masked_df: Any,
         strategy: str,
@@ -1001,7 +1846,7 @@ class TruthoundAdapter:
         """Convert truthound mask result to MaskResult.
 
         Args:
-            source: Original data source path.
+            source: Original data source (path or DataSource).
             output: Output file path.
             masked_df: Polars DataFrame with masked data.
             strategy: Masking strategy used.
@@ -1033,7 +1878,7 @@ class TruthoundAdapter:
             masked_df.write_csv(output)
 
         return MaskResult(
-            source=source,
+            source=_get_source_name(source),
             output_path=str(output_path.absolute()),
             row_count=row_count,
             column_count=len(all_columns),
@@ -1047,7 +1892,272 @@ class TruthoundAdapter:
         self._executor.shutdown(wait=False)
 
 
-# Singleton instance
+# =============================================================================
+# ValidationResult Mock for Reporter Integration
+# =============================================================================
+
+
+class _ValidationResultMock:
+    """Mock object that mimics truthound's ValidationResult interface.
+
+    This enables using truthound reporters with CheckResult objects from
+    this adapter, maintaining loose coupling with the truthound library.
+
+    The mock provides compatibility with truthound reporters that expect:
+    - ValidationResult from truthound.stores.results (new API)
+    - Report from truthound.report (legacy API)
+    """
+
+    def __init__(self, check_result: CheckResult) -> None:
+        from datetime import datetime
+
+        self._result = check_result
+        self._results = [
+            _ValidatorResultMock(issue) for issue in check_result.issues
+        ]
+        self._statistics = _ResultStatisticsMock(check_result)
+        self._run_time = check_result.run_time or datetime.now()
+
+    # === ValidationResult interface (new API) ===
+
+    @property
+    def run_id(self) -> str:
+        return self._result.run_id or f"run-{id(self._result)}"
+
+    @property
+    def run_time(self) -> Any:
+        return self._run_time
+
+    @property
+    def data_asset(self) -> str:
+        return self._result.source
+
+    @property
+    def status(self) -> "_ResultStatusMock":
+        return _ResultStatusMock(self._result.passed)
+
+    @property
+    def success(self) -> bool:
+        return self._result.passed
+
+    @property
+    def results(self) -> list["_ValidatorResultMock"]:
+        return self._results
+
+    @property
+    def statistics(self) -> "_ResultStatisticsMock":
+        return self._statistics
+
+    @property
+    def tags(self) -> dict[str, Any]:
+        return {}
+
+    # === Report interface (legacy API) ===
+
+    @property
+    def source(self) -> str:
+        return self._result.source
+
+    @property
+    def row_count(self) -> int:
+        return self._result.row_count
+
+    @property
+    def column_count(self) -> int:
+        return self._result.column_count
+
+    @property
+    def issues(self) -> list["_ValidatorResultMock"]:
+        return self._results
+
+    @property
+    def has_issues(self) -> bool:
+        return self._result.total_issues > 0
+
+    @property
+    def has_critical(self) -> bool:
+        return self._result.has_critical
+
+    @property
+    def has_high(self) -> bool:
+        return self._result.has_high
+
+    @property
+    def suite_name(self) -> str:
+        return "Truthound Validation"
+
+    def to_dict(self) -> dict[str, Any]:
+        return {
+            "run_id": self.run_id,
+            "run_time": (
+                self._run_time.isoformat()
+                if hasattr(self._run_time, "isoformat")
+                else str(self._run_time)
+            ),
+            "data_asset": self.data_asset,
+            "status": self.status.value,
+            "success": self.success,
+            "results": [r.to_dict() for r in self.results],
+            "statistics": self._statistics.to_dict(),
+        }
+
+    def to_json(self, indent: int | None = 2) -> str:
+        import json
+
+        return json.dumps(self.to_dict(), indent=indent, default=str)
+
+
+class _ResultStatusMock:
+    """Mock ResultStatus enum for reporter compatibility."""
+
+    def __init__(self, passed: bool) -> None:
+        self._passed = passed
+
+    @property
+    def value(self) -> str:
+        return "SUCCESS" if self._passed else "FAILURE"
+
+    def __str__(self) -> str:
+        return self.value
+
+
+class _ResultStatisticsMock:
+    """Mock ResultStatistics for reporter compatibility."""
+
+    def __init__(self, check_result: CheckResult) -> None:
+        self._result = check_result
+
+    @property
+    def total_issues(self) -> int:
+        return self._result.total_issues
+
+    @property
+    def total_rows(self) -> int:
+        return self._result.row_count
+
+    @property
+    def total_columns(self) -> int:
+        return self._result.column_count
+
+    @property
+    def critical_count(self) -> int:
+        return self._result.critical_issues
+
+    @property
+    def high_count(self) -> int:
+        return self._result.high_issues
+
+    @property
+    def medium_count(self) -> int:
+        return self._result.medium_issues
+
+    @property
+    def low_count(self) -> int:
+        return self._result.low_issues
+
+    @property
+    def passed(self) -> bool:
+        return self._result.passed
+
+    def to_dict(self) -> dict[str, Any]:
+        return {
+            "total_issues": self.total_issues,
+            "total_rows": self.total_rows,
+            "total_columns": self.total_columns,
+            "critical_count": self.critical_count,
+            "high_count": self.high_count,
+            "medium_count": self.medium_count,
+            "low_count": self.low_count,
+            "passed": self.passed,
+        }
+
+
+class _ValidatorResultMock:
+    """Mock ValidatorResult for reporter compatibility."""
+
+    def __init__(self, issue: dict[str, Any]) -> None:
+        self._issue = issue
+
+    @property
+    def validator_name(self) -> str:
+        return self._issue.get("validator_name") or self._issue.get("issue_type", "")
+
+    @property
+    def column(self) -> str | None:
+        return self._issue.get("column")
+
+    @property
+    def issue_type(self) -> str:
+        return self._issue.get("issue_type", "")
+
+    @property
+    def severity(self) -> "_SeverityMock":
+        return _SeverityMock(self._issue.get("severity", "medium"))
+
+    @property
+    def message(self) -> str:
+        return self._issue.get("message", "")
+
+    @property
+    def count(self) -> int:
+        return self._issue.get("count", 0)
+
+    @property
+    def success(self) -> bool:
+        return False  # All issues are failures
+
+    @property
+    def expected(self) -> Any:
+        return self._issue.get("expected")
+
+    @property
+    def actual(self) -> Any:
+        return self._issue.get("actual")
+
+    @property
+    def details(self) -> dict[str, Any]:
+        return self._issue.get("details") or {}
+
+    @property
+    def sample_values(self) -> list[Any]:
+        return self._issue.get("sample_values") or []
+
+    def to_dict(self) -> dict[str, Any]:
+        return {
+            "validator_name": self.validator_name,
+            "column": self.column,
+            "issue_type": self.issue_type,
+            "severity": self.severity.value,
+            "message": self.message,
+            "count": self.count,
+            "success": self.success,
+            "expected": self.expected,
+            "actual": self.actual,
+            "details": self.details,
+            "sample_values": self.sample_values,
+        }
+
+
+class _SeverityMock:
+    """Mock Severity enum for reporter compatibility."""
+
+    def __init__(self, value: str) -> None:
+        self._value = value.lower() if isinstance(value, str) else str(value).lower()
+
+    @property
+    def value(self) -> str:
+        return self._value
+
+    def __str__(self) -> str:
+        return self._value
+
+
+# =============================================================================
+# Singleton Management
+# =============================================================================
+
+
+# Singleton instance
 _adapter: TruthoundAdapter | None = None
 
 
@@ -1072,3 +2182,1316 @@ def reset_adapter() -> None:
     if _adapter is not None:
         _adapter.shutdown()
         _adapter = None
+
+
+# =============================================================================
+# Schema Evolution API (truthound.profiler.evolution)
+# =============================================================================
+
+
+@dataclass
+class SchemaChangeResult:
+    """Schema change detection result.
+
+    Represents a single detected change between schema versions.
+
+    Attributes:
+        change_type: Type of change (column_added, column_removed, type_changed, etc.)
+        column_name: Name of the affected column.
+        old_value: Previous value (type, nullable, etc.)
+        new_value: New value.
+        severity: Change severity (info, warning, critical).
+        breaking: Whether this is a breaking change.
+        description: Human-readable description.
+        migration_hint: Suggestion for handling the change.
+    """
+
+    change_type: str
+    column_name: str
+    old_value: Any
+    new_value: Any
+    severity: str
+    breaking: bool
+    description: str
+    migration_hint: str | None = None
+
+    def to_dict(self) -> dict[str, Any]:
+        """Convert to dictionary."""
+        return {
+            "change_type": self.change_type,
+            "column_name": self.column_name,
+            "old_value": self.old_value,
+            "new_value": self.new_value,
+            "severity": self.severity,
+            "breaking": self.breaking,
+            "description": self.description,
+            "migration_hint": self.migration_hint,
+        }
+
+
+@dataclass
+class SchemaDetectionResult:
+    """Schema evolution detection result.
+
+    Result from comparing two schemas.
+
+    Attributes:
+        total_changes: Total number of changes detected.
+        breaking_changes: Number of breaking changes.
+        compatibility_level: Compatibility assessment (compatible, minor, breaking).
+        changes: List of individual changes.
+    """
+
+    total_changes: int
+    breaking_changes: int
+    compatibility_level: str
+    changes: list[SchemaChangeResult]
+
+    def to_dict(self) -> dict[str, Any]:
+        """Convert to dictionary."""
+        return {
+            "total_changes": self.total_changes,
+            "breaking_changes": self.breaking_changes,
+            "compatibility_level": self.compatibility_level,
+            "changes": [c.to_dict() for c in self.changes],
+        }
+
+
+@dataclass
+class RenameDetectionResult:
+    """Column rename detection result.
+
+    Attributes:
+        old_name: Original column name.
+        new_name: New column name.
+        similarity: Similarity score (0.0-1.0).
+        confidence: Confidence level (high, medium, low).
+        reasons: Reasons for the rename detection.
+    """
+
+    old_name: str
+    new_name: str
+    similarity: float
+    confidence: str
+    reasons: list[str]
+
+    def to_dict(self) -> dict[str, Any]:
+        """Convert to dictionary."""
+        return {
+            "old_name": self.old_name,
+            "new_name": self.new_name,
+            "similarity": self.similarity,
+            "confidence": self.confidence,
+            "reasons": self.reasons,
+        }
+
+
+@dataclass
+class RenameDetectionSummary:
+    """Summary of rename detection results.
+
+    Attributes:
+        confirmed_renames: High-confidence confirmed renames.
+        possible_renames: Lower-confidence possible renames.
+        unmatched_added: Columns added without rename match.
+        unmatched_removed: Columns removed without rename match.
+    """
+
+    confirmed_renames: list[RenameDetectionResult]
+    possible_renames: list[RenameDetectionResult]
+    unmatched_added: list[str]
+    unmatched_removed: list[str]
+
+    def to_dict(self) -> dict[str, Any]:
+        """Convert to dictionary."""
+        return {
+            "confirmed_renames": [r.to_dict() for r in self.confirmed_renames],
+            "possible_renames": [r.to_dict() for r in self.possible_renames],
+            "unmatched_added": self.unmatched_added,
+            "unmatched_removed": self.unmatched_removed,
+        }
+
+
+@dataclass
+class SchemaVersionResult:
+    """Schema version information.
+
+    Attributes:
+        id: Version identifier (hash or version string).
+        version: Version string (e.g., "1.0.0", "20260129.143000").
+        schema: Schema dictionary.
+        metadata: Optional metadata.
+        created_at: Creation timestamp.
+        has_breaking_changes: Whether this version has breaking changes from parent.
+        changes_from_parent: List of changes from parent version.
+    """
+
+    id: str
+    version: str
+    schema: dict[str, Any]
+    metadata: dict[str, Any] | None
+    created_at: str | None
+    has_breaking_changes: bool = False
+    changes_from_parent: list[SchemaChangeResult] | None = None
+
+    def to_dict(self) -> dict[str, Any]:
+        """Convert to dictionary."""
+        return {
+            "id": self.id,
+            "version": self.version,
+            "schema": self.schema,
+            "metadata": self.metadata,
+            "created_at": self.created_at,
+            "has_breaking_changes": self.has_breaking_changes,
+            "changes_from_parent": (
+                [c.to_dict() for c in self.changes_from_parent]
+                if self.changes_from_parent
+                else None
+            ),
+        }
+
+
+@dataclass
+class SchemaDiffResult:
+    """Schema diff between two versions.
+
+    Attributes:
+        from_version: Source version string.
+        to_version: Target version string.
+        changes: List of changes.
+        text_diff: Human-readable text diff.
+    """
+
+    from_version: str
+    to_version: str
+    changes: list[SchemaChangeResult]
+    text_diff: str
+
+    def to_dict(self) -> dict[str, Any]:
+        """Convert to dictionary."""
+        return {
+            "from_version": self.from_version,
+            "to_version": self.to_version,
+            "changes": [c.to_dict() for c in self.changes],
+            "text_diff": self.text_diff,
+        }
+
+
+@dataclass
+class SchemaWatcherEvent:
+    """Schema watcher change event.
+
+    Attributes:
+        source: Source name that changed.
+        has_breaking_changes: Whether breaking changes were detected.
+        total_changes: Total number of changes.
+        changes: List of changes.
+        timestamp: Event timestamp.
+    """
+
+    source: str
+    has_breaking_changes: bool
+    total_changes: int
+    changes: list[SchemaChangeResult]
+    timestamp: str
+
+    def to_dict(self) -> dict[str, Any]:
+        """Convert to dictionary."""
+        return {
+            "source": self.source,
+            "has_breaking_changes": self.has_breaking_changes,
+            "total_changes": self.total_changes,
+            "changes": [c.to_dict() for c in self.changes],
+            "timestamp": self.timestamp,
+        }
+
+
+@dataclass
+class BreakingChangeAlert:
+    """Breaking change alert with impact analysis.
+
+    Attributes:
+        alert_id: Unique alert identifier.
+        title: Alert title.
+        source: Source name.
+        changes: List of breaking changes.
+        impact_scope: Impact scope (local, downstream, system).
+        affected_consumers: List of affected consumers.
+        data_risk_level: Risk level (1-5).
+        recommendations: List of recommendations.
+        status: Alert status (open, acknowledged, resolved).
+        created_at: Creation timestamp.
+        acknowledged_at: Acknowledgment timestamp.
+        resolved_at: Resolution timestamp.
+    """
+
+    alert_id: str
+    title: str
+    source: str
+    changes: list[SchemaChangeResult]
+    impact_scope: str
+    affected_consumers: list[str]
+    data_risk_level: int
+    recommendations: list[str]
+    status: str
+    created_at: str
+    acknowledged_at: str | None = None
+    resolved_at: str | None = None
+
+    def to_dict(self) -> dict[str, Any]:
+        """Convert to dictionary."""
+        return {
+            "alert_id": self.alert_id,
+            "title": self.title,
+            "source": self.source,
+            "changes": [c.to_dict() for c in self.changes],
+            "impact_scope": self.impact_scope,
+            "affected_consumers": self.affected_consumers,
+            "data_risk_level": self.data_risk_level,
+            "recommendations": self.recommendations,
+            "status": self.status,
+            "created_at": self.created_at,
+            "acknowledged_at": self.acknowledged_at,
+            "resolved_at": self.resolved_at,
+        }
+
+
+class SchemaEvolutionAdapter:
+    """Async wrapper for truthound schema evolution functions.
+
+    This adapter provides an async interface to truthound's schema evolution
+    module (truthound.profiler.evolution), including:
+    - SchemaEvolutionDetector for change detection
+    - SchemaHistory for version management
+    - SchemaWatcher for continuous monitoring
+    - ColumnRenameDetector for rename detection
+    - BreakingChangeAlertManager for alert management
+    - ImpactAnalyzer for impact analysis
+
+    All operations run in a thread pool to avoid blocking the event loop.
+    """
+
+    def __init__(self, max_workers: int = 4) -> None:
+        """Initialize adapter.
+
+        Args:
+            max_workers: Maximum worker threads for concurrent operations.
+        """
+        self._executor = ThreadPoolExecutor(max_workers=max_workers)
+        self._watchers: dict[str, Any] = {}  # watcher_id -> SchemaWatcher
+        self._histories: dict[str, Any] = {}  # history_id -> SchemaHistory
+        self._alert_manager: Any = None
+        self._impact_analyzer: Any = None
+
+    async def detect_changes(
+        self,
+        current_schema: dict[str, Any],
+        baseline_schema: dict[str, Any],
+        *,
+        detect_renames: bool = True,
+        rename_similarity_threshold: float = 0.8,
+    ) -> SchemaDetectionResult:
+        """Detect schema changes between two schemas.
+
+        Uses truthound's SchemaEvolutionDetector for comprehensive change
+        detection including column additions, removals, type changes, and renames.
+
+        Args:
+            current_schema: Current schema dictionary ({"column": "Type"}).
+            baseline_schema: Baseline schema dictionary.
+            detect_renames: Enable rename detection.
+            rename_similarity_threshold: Threshold for considering a rename (0.0-1.0).
+
+        Returns:
+            SchemaDetectionResult with all detected changes.
+        """
+        from truthound.profiler.evolution import SchemaEvolutionDetector
+
+        def _detect():
+            detector = SchemaEvolutionDetector(
+                detect_renames=detect_renames,
+                rename_similarity_threshold=rename_similarity_threshold,
+            )
+            changes = detector.detect_changes(current_schema, baseline_schema)
+            summary = detector.get_change_summary(changes)
+            return changes, summary
+
+        loop = asyncio.get_event_loop()
+        changes, summary = await loop.run_in_executor(self._executor, _detect)
+
+        return self._convert_detection_result(changes, summary)
+
+    async def detect_renames(
+        self,
+        added_columns: dict[str, str],
+        removed_columns: dict[str, str],
+        *,
+        similarity_threshold: float = 0.8,
+        require_type_match: bool = True,
+        allow_compatible_types: bool = True,
+        algorithm: str = "composite",
+    ) -> RenameDetectionSummary:
+        """Detect column renames between added and removed columns.
+
+        Uses truthound's ColumnRenameDetector with configurable similarity
+        algorithms for accurate rename detection.
+
+        Args:
+            added_columns: Dict of added columns {"name": "Type"}.
+            removed_columns: Dict of removed columns {"name": "Type"}.
+            similarity_threshold: Threshold for considering a rename (0.0-1.0).
+            require_type_match: Require matching types for rename.
+            allow_compatible_types: Allow compatible type changes (e.g., Int32->Int64).
+            algorithm: Similarity algorithm:
+                - "composite": Weighted combination (default)
+                - "levenshtein": Edit distance
+                - "jaro_winkler": Short strings, prefixes
+                - "ngram": Partial matches
+                - "token": snake_case/camelCase names
+
+        Returns:
+            RenameDetectionSummary with confirmed and possible renames.
+        """
+        from truthound.profiler.evolution import ColumnRenameDetector
+
+        def _detect():
+            detector = ColumnRenameDetector(
+                similarity_threshold=similarity_threshold,
+                require_type_match=require_type_match,
+                allow_compatible_types=allow_compatible_types,
+            )
+            return detector.detect(
+                added_columns=added_columns,
+                removed_columns=removed_columns,
+            )
+
+        loop = asyncio.get_event_loop()
+        result = await loop.run_in_executor(self._executor, _detect)
+
+        return self._convert_rename_result(result)
+
+    async def create_history(
+        self,
+        history_id: str,
+        storage_path: str,
+        *,
+        version_strategy: str = "semantic",
+        max_versions: int = 100,
+        compress: bool = True,
+    ) -> str:
+        """Create a new schema history storage.
+
+        Uses truthound's SchemaHistory for version management with support
+        for semantic, incremental, timestamp, and git versioning strategies.
+
+        Args:
+            history_id: Unique identifier for this history instance.
+            storage_path: Path for file-based storage.
+            version_strategy: Version numbering strategy:
+                - "semantic": 1.2.3 format, auto-bumps based on change type
+                - "incremental": 1, 2, 3 simple numbers
+                - "timestamp": 20260128.143052 time-based
+                - "git": a1b2c3d4 git-like hashes
+            max_versions: Maximum versions to keep.
+            compress: Compress stored files.
+
+        Returns:
+            History ID for future operations.
+        """
+        from truthound.profiler.evolution import SchemaHistory
+
+        def _create():
+            return SchemaHistory.create(
+                storage_type="file",
+                path=storage_path,
+                version_strategy=version_strategy,
+                max_versions=max_versions,
+                compress=compress,
+            )
+
+        loop = asyncio.get_event_loop()
+        history = await loop.run_in_executor(self._executor, _create)
+
+        self._histories[history_id] = history
+        return history_id
+
+    async def save_schema_version(
+        self,
+        history_id: str,
+        schema: dict[str, Any],
+        *,
+        version: str | None = None,
+        metadata: dict[str, Any] | None = None,
+    ) -> SchemaVersionResult:
+        """Save a schema version to history.
+
+        Args:
+            history_id: History instance ID.
+            schema: Schema dictionary to save.
+            version: Optional explicit version string.
+            metadata: Optional metadata (author, message, etc.).
+
+        Returns:
+            SchemaVersionResult with version info.
+
+        Raises:
+            ValueError: If history_id not found.
+        """
+        if history_id not in self._histories:
+            raise ValueError(f"History '{history_id}' not found")
+
+        history = self._histories[history_id]
+
+        def _save():
+            kwargs: dict[str, Any] = {}
+            if version:
+                kwargs["version"] = version
+            if metadata:
+                kwargs["metadata"] = metadata
+            return history.save(schema, **kwargs)
+
+        loop = asyncio.get_event_loop()
+        result = await loop.run_in_executor(self._executor, _save)
+
+        return self._convert_version_result(result)
+
+    async def get_schema_version(
+        self,
+        history_id: str,
+        version: str,
+    ) -> SchemaVersionResult | None:
+        """Get a specific schema version.
+
+        Args:
+            history_id: History instance ID.
+            version: Version string or ID.
+
+        Returns:
+            SchemaVersionResult or None if not found.
+        """
+        if history_id not in self._histories:
+            raise ValueError(f"History '{history_id}' not found")
+
+        history = self._histories[history_id]
+
+        def _get():
+            try:
+                return history.get_by_version(version)
+            except Exception:
+                return history.get(version)
+
+        loop = asyncio.get_event_loop()
+        result = await loop.run_in_executor(self._executor, _get)
+
+        if result is None:
+            return None
+        return self._convert_version_result(result)
+
+    async def list_schema_versions(
+        self,
+        history_id: str,
+        *,
+        limit: int = 50,
+        since: str | None = None,
+    ) -> list[SchemaVersionResult]:
+        """List schema versions in history.
+
+        Args:
+            history_id: History instance ID.
+            limit: Maximum versions to return.
+            since: Filter versions since this datetime (ISO format).
+
+        Returns:
+            List of SchemaVersionResult.
+        """
+        from datetime import datetime, timedelta
+
+        if history_id not in self._histories:
+            raise ValueError(f"History '{history_id}' not found")
+
+        history = self._histories[history_id]
+
+        def _list():
+            kwargs: dict[str, Any] = {"limit": limit}
+            if since:
+                kwargs["since"] = datetime.fromisoformat(since)
+            return history.list(**kwargs)
+
+        loop = asyncio.get_event_loop()
+        versions = await loop.run_in_executor(self._executor, _list)
+
+        return [self._convert_version_result(v) for v in versions]
+
+    async def get_latest_version(
+        self,
+        history_id: str,
+    ) -> SchemaVersionResult | None:
+        """Get the latest schema version.
+
+        Args:
+            history_id: History instance ID.
+
+        Returns:
+            Latest SchemaVersionResult or None.
+        """
+        if history_id not in self._histories:
+            raise ValueError(f"History '{history_id}' not found")
+
+        history = self._histories[history_id]
+
+        loop = asyncio.get_event_loop()
+        result = await loop.run_in_executor(
+            self._executor, lambda: history.latest
+        )
+
+        if result is None:
+            return None
+        return self._convert_version_result(result)
+
+    async def diff_versions(
+        self,
+        history_id: str,
+        from_version: str,
+        to_version: str | None = None,
+    ) -> SchemaDiffResult:
+        """Get diff between two schema versions.
+
+        Args:
+            history_id: History instance ID.
+            from_version: Source version string.
+            to_version: Target version string (None = latest).
+
+        Returns:
+            SchemaDiffResult with changes and text diff.
+        """
+        if history_id not in self._histories:
+            raise ValueError(f"History '{history_id}' not found")
+
+        history = self._histories[history_id]
+
+        def _diff():
+            if to_version:
+                return history.diff(from_version, to_version)
+            else:
+                return history.diff(from_version)
+
+        loop = asyncio.get_event_loop()
+        diff = await loop.run_in_executor(self._executor, _diff)
+
+        return self._convert_diff_result(diff, from_version, to_version or "latest")
+
+    async def has_breaking_changes_since(
+        self,
+        history_id: str,
+        version: str,
+    ) -> bool:
+        """Check if there are breaking changes since a version.
+
+        Args:
+            history_id: History instance ID.
+            version: Version to check from.
+
+        Returns:
+            True if breaking changes exist.
+        """
+        if history_id not in self._histories:
+            raise ValueError(f"History '{history_id}' not found")
+
+        history = self._histories[history_id]
+
+        loop = asyncio.get_event_loop()
+        return await loop.run_in_executor(
+            self._executor, lambda: history.has_breaking_changes_since(version)
+        )
+
+    async def rollback_version(
+        self,
+        history_id: str,
+        to_version: str,
+        *,
+        reason: str | None = None,
+    ) -> SchemaVersionResult:
+        """Rollback to a previous version.
+
+        Creates a new version that matches the specified version.
+
+        Args:
+            history_id: History instance ID.
+            to_version: Version to rollback to.
+            reason: Reason for rollback.
+
+        Returns:
+            New SchemaVersionResult after rollback.
+        """
+        if history_id not in self._histories:
+            raise ValueError(f"History '{history_id}' not found")
+
+        history = self._histories[history_id]
+
+        def _rollback():
+            kwargs: dict[str, Any] = {}
+            if reason:
+                kwargs["reason"] = reason
+            return history.rollback(to_version, **kwargs)
+
+        loop = asyncio.get_event_loop()
+        result = await loop.run_in_executor(self._executor, _rollback)
+
+        return self._convert_version_result(result)
+
+    async def create_watcher(
+        self,
+        watcher_id: str,
+        sources: list[dict[str, Any]],
+        *,
+        poll_interval: int = 60,
+        only_breaking: bool = False,
+        enable_history: bool = True,
+        history_path: str | None = None,
+    ) -> str:
+        """Create a new schema watcher.
+
+        Uses truthound's SchemaWatcher for continuous monitoring with
+        configurable sources, handlers, and polling.
+
+        Args:
+            watcher_id: Unique identifier for this watcher.
+            sources: List of source configurations, each with:
+                - type: "file", "dict", or "polars"
+                - path: For file sources
+                - schema: For dict sources
+                - name: Source name
+            poll_interval: Polling interval in seconds.
+            only_breaking: Only alert on breaking changes.
+            enable_history: Enable history tracking.
+            history_path: Path for history storage.
+
+        Returns:
+            Watcher ID for future operations.
+        """
+        from truthound.profiler.evolution import (
+            SchemaWatcher,
+            FileSchemaSource,
+            DictSchemaSource,
+            LoggingEventHandler,
+            HistoryEventHandler,
+            SchemaHistory,
+        )
+
+        def _create():
+            watcher = SchemaWatcher()
+
+            # Add sources
+            for src in sources:
+                src_type = src.get("type", "file")
+                if src_type == "file":
+                    watcher.add_source(FileSchemaSource(src["path"]))
+                elif src_type == "dict":
+                    watcher.add_source(
+                        DictSchemaSource(src["schema"], src.get("name", "dict"))
+                    )
+
+            # Add logging handler
+            watcher.add_handler(LoggingEventHandler())
+
+            # Add history handler if enabled
+            if enable_history and history_path:
+                history = SchemaHistory.create(
+                    storage_type="file",
+                    path=history_path,
+                )
+                watcher.add_handler(HistoryEventHandler(history))
+
+            return watcher
+
+        loop = asyncio.get_event_loop()
+        watcher = await loop.run_in_executor(self._executor, _create)
+
+        self._watchers[watcher_id] = {
+            "watcher": watcher,
+            "poll_interval": poll_interval,
+            "only_breaking": only_breaking,
+            "status": "created",
+        }
+        return watcher_id
+
+    async def start_watcher(
+        self,
+        watcher_id: str,
+        *,
+        daemon: bool = True,
+    ) -> None:
+        """Start a schema watcher.
+
+        Args:
+            watcher_id: Watcher ID to start.
+            daemon: Run as daemon thread.
+
+        Raises:
+            ValueError: If watcher_id not found.
+        """
+        if watcher_id not in self._watchers:
+            raise ValueError(f"Watcher '{watcher_id}' not found")
+
+        watcher_data = self._watchers[watcher_id]
+        watcher = watcher_data["watcher"]
+        poll_interval = watcher_data["poll_interval"]
+
+        def _start():
+            watcher.start(poll_interval=poll_interval, daemon=daemon)
+
+        loop = asyncio.get_event_loop()
+        await loop.run_in_executor(self._executor, _start)
+
+        watcher_data["status"] = "running"
+
+    async def stop_watcher(self, watcher_id: str) -> None:
+        """Stop a schema watcher.
+
+        Args:
+            watcher_id: Watcher ID to stop.
+        """
+        if watcher_id not in self._watchers:
+            raise ValueError(f"Watcher '{watcher_id}' not found")
+
+        watcher_data = self._watchers[watcher_id]
+        watcher = watcher_data["watcher"]
+
+        loop = asyncio.get_event_loop()
+        await loop.run_in_executor(self._executor, watcher.stop)
+
+        watcher_data["status"] = "stopped"
+
+    async def pause_watcher(self, watcher_id: str) -> None:
+        """Pause a schema watcher.
+
+        Args:
+            watcher_id: Watcher ID to pause.
+        """
+        if watcher_id not in self._watchers:
+            raise ValueError(f"Watcher '{watcher_id}' not found")
+
+        watcher_data = self._watchers[watcher_id]
+        watcher = watcher_data["watcher"]
+
+        loop = asyncio.get_event_loop()
+        await loop.run_in_executor(self._executor, watcher.pause)
+
+        watcher_data["status"] = "paused"
+
+    async def resume_watcher(self, watcher_id: str) -> None:
+        """Resume a paused schema watcher.
+
+        Args:
+            watcher_id: Watcher ID to resume.
+        """
+        if watcher_id not in self._watchers:
+            raise ValueError(f"Watcher '{watcher_id}' not found")
+
+        watcher_data = self._watchers[watcher_id]
+        watcher = watcher_data["watcher"]
+
+        loop = asyncio.get_event_loop()
+        await loop.run_in_executor(self._executor, watcher.resume)
+
+        watcher_data["status"] = "running"
+
+    async def check_watcher_now(
+        self,
+        watcher_id: str,
+    ) -> list[SchemaWatcherEvent]:
+        """Execute immediate check for a watcher.
+
+        Args:
+            watcher_id: Watcher ID to check.
+
+        Returns:
+            List of SchemaWatcherEvent for any detected changes.
+        """
+        if watcher_id not in self._watchers:
+            raise ValueError(f"Watcher '{watcher_id}' not found")
+
+        watcher_data = self._watchers[watcher_id]
+        watcher = watcher_data["watcher"]
+
+        loop = asyncio.get_event_loop()
+        events = await loop.run_in_executor(self._executor, watcher.check_now)
+
+        return [self._convert_watcher_event(e) for e in events]
+
+    async def get_watcher_status(self, watcher_id: str) -> dict[str, Any]:
+        """Get watcher status.
+
+        Args:
+            watcher_id: Watcher ID.
+
+        Returns:
+            Status dictionary with status, poll_interval, only_breaking.
+        """
+        if watcher_id not in self._watchers:
+            raise ValueError(f"Watcher '{watcher_id}' not found")
+
+        watcher_data = self._watchers[watcher_id]
+        return {
+            "watcher_id": watcher_id,
+            "status": watcher_data["status"],
+            "poll_interval": watcher_data["poll_interval"],
+            "only_breaking": watcher_data["only_breaking"],
+        }
+
+    async def delete_watcher(self, watcher_id: str) -> None:
+        """Delete a watcher.
+
+        Stops the watcher if running and removes it.
+
+        Args:
+            watcher_id: Watcher ID to delete.
+        """
+        if watcher_id not in self._watchers:
+            raise ValueError(f"Watcher '{watcher_id}' not found")
+
+        watcher_data = self._watchers[watcher_id]
+        if watcher_data["status"] == "running":
+            await self.stop_watcher(watcher_id)
+
+        del self._watchers[watcher_id]
+
+    async def setup_impact_analyzer(
+        self,
+        consumers: dict[str, list[str]] | None = None,
+        queries: dict[str, list[str]] | None = None,
+    ) -> None:
+        """Setup impact analyzer with consumer mappings.
+
+        Args:
+            consumers: Dict of consumer name -> list of sources it depends on.
+            queries: Dict of source name -> list of queries using it.
+        """
+        from truthound.profiler.evolution import ImpactAnalyzer
+
+        def _setup():
+            analyzer = ImpactAnalyzer()
+            if consumers:
+                for consumer, sources in consumers.items():
+                    analyzer.register_consumer(consumer, sources)
+            if queries:
+                for source, query_list in queries.items():
+                    for query in query_list:
+                        analyzer.register_query(source, query)
+            return analyzer
+
+        loop = asyncio.get_event_loop()
+        self._impact_analyzer = await loop.run_in_executor(self._executor, _setup)
+
+    async def setup_alert_manager(
+        self,
+        alert_storage_path: str,
+    ) -> None:
+        """Setup breaking change alert manager.
+
+        Args:
+            alert_storage_path: Path for alert storage.
+        """
+        from truthound.profiler.evolution import BreakingChangeAlertManager
+
+        def _setup():
+            return BreakingChangeAlertManager(
+                impact_analyzer=self._impact_analyzer,
+                alert_storage_path=alert_storage_path,
+            )
+
+        loop = asyncio.get_event_loop()
+        self._alert_manager = await loop.run_in_executor(self._executor, _setup)
+
+    async def create_alert(
+        self,
+        changes: list[dict[str, Any]],
+        source: str,
+    ) -> BreakingChangeAlert:
+        """Create a breaking change alert.
+
+        Args:
+            changes: List of change dictionaries from detect_changes.
+            source: Source name.
+
+        Returns:
+            BreakingChangeAlert with impact analysis.
+
+        Raises:
+            ValueError: If alert manager not setup.
+        """
+        if self._alert_manager is None:
+            raise ValueError("Alert manager not setup. Call setup_alert_manager first.")
+
+        def _create():
+            return self._alert_manager.create_alert(changes, source=source)
+
+        loop = asyncio.get_event_loop()
+        alert = await loop.run_in_executor(self._executor, _create)
+
+        return self._convert_alert_result(alert)
+
+    async def acknowledge_alert(self, alert_id: str) -> None:
+        """Acknowledge an alert.
+
+        Args:
+            alert_id: Alert ID to acknowledge.
+        """
+        if self._alert_manager is None:
+            raise ValueError("Alert manager not setup.")
+
+        loop = asyncio.get_event_loop()
+        await loop.run_in_executor(
+            self._executor, lambda: self._alert_manager.acknowledge_alert(alert_id)
+        )
+
+    async def resolve_alert(self, alert_id: str) -> None:
+        """Resolve an alert.
+
+        Args:
+            alert_id: Alert ID to resolve.
+        """
+        if self._alert_manager is None:
+            raise ValueError("Alert manager not setup.")
+
+        loop = asyncio.get_event_loop()
+        await loop.run_in_executor(
+            self._executor, lambda: self._alert_manager.resolve_alert(alert_id)
+        )
+
+    async def get_alert_history(
+        self,
+        *,
+        status: str | None = None,
+    ) -> list[BreakingChangeAlert]:
+        """Get alert history.
+
+        Args:
+            status: Filter by status (open, acknowledged, resolved).
+
+        Returns:
+            List of BreakingChangeAlert.
+        """
+        if self._alert_manager is None:
+            raise ValueError("Alert manager not setup.")
+
+        def _get():
+            kwargs: dict[str, Any] = {}
+            if status:
+                kwargs["status"] = status
+            return self._alert_manager.get_alert_history(**kwargs)
+
+        loop = asyncio.get_event_loop()
+        alerts = await loop.run_in_executor(self._executor, _get)
+
+        return [self._convert_alert_result(a) for a in alerts]
+
+    async def get_alert_stats(self) -> dict[str, int]:
+        """Get alert statistics.
+
+        Returns:
+            Dict with total, open, acknowledged, resolved counts.
+        """
+        if self._alert_manager is None:
+            raise ValueError("Alert manager not setup.")
+
+        loop = asyncio.get_event_loop()
+        return await loop.run_in_executor(
+            self._executor, self._alert_manager.get_stats
+        )
+
+    # =========================================================================
+    # Result Conversion Methods
+    # =========================================================================
+
+    def _convert_detection_result(
+        self,
+        changes: list[Any],
+        summary: Any,
+    ) -> SchemaDetectionResult:
+        """Convert truthound detection result."""
+        converted_changes = []
+        for c in changes:
+            converted_changes.append(
+                SchemaChangeResult(
+                    change_type=c.change_type.value if hasattr(c.change_type, "value") else str(c.change_type),
+                    column_name=getattr(c, "column", getattr(c, "column_name", "")),
+                    old_value=getattr(c, "old_value", None),
+                    new_value=getattr(c, "new_value", None),
+                    severity=c.severity.value if hasattr(c.severity, "value") else str(c.severity),
+                    breaking=getattr(c, "breaking", False),
+                    description=getattr(c, "description", ""),
+                    migration_hint=getattr(c, "migration_hint", None),
+                )
+            )
+
+        compatibility = "compatible"
+        if hasattr(summary, "compatibility_level"):
+            compatibility = (
+                summary.compatibility_level.value
+                if hasattr(summary.compatibility_level, "value")
+                else str(summary.compatibility_level)
+            )
+
+        return SchemaDetectionResult(
+            total_changes=getattr(summary, "total_changes", len(changes)),
+            breaking_changes=getattr(summary, "breaking_changes", 0),
+            compatibility_level=compatibility,
+            changes=converted_changes,
+        )
+
+    def _convert_rename_result(self, result: Any) -> RenameDetectionSummary:
+        """Convert truthound rename detection result."""
+        confirmed = []
+        for r in getattr(result, "confirmed_renames", []):
+            confirmed.append(
+                RenameDetectionResult(
+                    old_name=r.old_name,
+                    new_name=r.new_name,
+                    similarity=r.similarity,
+                    confidence=r.confidence.value if hasattr(r.confidence, "value") else str(r.confidence),
+                    reasons=list(getattr(r, "reasons", [])),
+                )
+            )
+
+        possible = []
+        for r in getattr(result, "possible_renames", []):
+            possible.append(
+                RenameDetectionResult(
+                    old_name=r.old_name,
+                    new_name=r.new_name,
+                    similarity=r.similarity,
+                    confidence=r.confidence.value if hasattr(r.confidence, "value") else str(r.confidence),
+                    reasons=list(getattr(r, "reasons", [])),
+                )
+            )
+
+        return RenameDetectionSummary(
+            confirmed_renames=confirmed,
+            possible_renames=possible,
+            unmatched_added=list(getattr(result, "unmatched_added", [])),
+            unmatched_removed=list(getattr(result, "unmatched_removed", [])),
+        )
+
+    def _convert_version_result(self, result: Any) -> SchemaVersionResult:
+        """Convert truthound version result."""
+        from datetime import datetime
+
+        created_at = None
+        if hasattr(result, "created_at") and result.created_at:
+            created_at = (
+                result.created_at.isoformat()
+                if isinstance(result.created_at, datetime)
+                else str(result.created_at)
+            )
+
+        changes = None
+        if hasattr(result, "changes_from_parent") and result.changes_from_parent:
+            changes = [
+                SchemaChangeResult(
+                    change_type=c.change_type.value if hasattr(c.change_type, "value") else str(c.change_type),
+                    column_name=getattr(c, "column", getattr(c, "column_name", "")),
+                    old_value=getattr(c, "old_value", None),
+                    new_value=getattr(c, "new_value", None),
+                    severity=c.severity.value if hasattr(c.severity, "value") else str(c.severity),
+                    breaking=getattr(c, "breaking", False),
+                    description=getattr(c, "description", ""),
+                    migration_hint=getattr(c, "migration_hint", None),
+                )
+                for c in result.changes_from_parent
+            ]
+
+        # Get schema as dict
+        schema = {}
+        if hasattr(result, "schema"):
+            schema = result.schema if isinstance(result.schema, dict) else {}
+        elif hasattr(result, "to_dict"):
+            schema = result.to_dict().get("schema", {})
+
+        return SchemaVersionResult(
+            id=getattr(result, "id", getattr(result, "version_id", "")),
+            version=str(getattr(result, "version", "")),
+            schema=schema,
+            metadata=getattr(result, "metadata", None),
+            created_at=created_at,
+            has_breaking_changes=getattr(result, "has_breaking_changes", False),
+            changes_from_parent=changes,
+        )
+
+    def _convert_diff_result(
+        self,
+        diff: Any,
+        from_version: str,
+        to_version: str,
+    ) -> SchemaDiffResult:
+        """Convert truthound diff result."""
+        changes = []
+        for c in getattr(diff, "changes", []):
+            changes.append(
+                SchemaChangeResult(
+                    change_type=c.change_type.value if hasattr(c.change_type, "value") else str(c.change_type),
+                    column_name=getattr(c, "column", getattr(c, "column_name", "")),
+                    old_value=getattr(c, "old_value", None),
+                    new_value=getattr(c, "new_value", None),
+                    severity=c.severity.value if hasattr(c.severity, "value") else str(c.severity),
+                    breaking=getattr(c, "breaking", False),
+                    description=getattr(c, "description", ""),
+                    migration_hint=getattr(c, "migration_hint", None),
+                )
+            )
+
+        text_diff = ""
+        if hasattr(diff, "format_text"):
+            text_diff = diff.format_text()
+
+        return SchemaDiffResult(
+            from_version=from_version,
+            to_version=to_version,
+            changes=changes,
+            text_diff=text_diff,
+        )
+
+    def _convert_watcher_event(self, event: Any) -> SchemaWatcherEvent:
+        """Convert truthound watcher event."""
+        from datetime import datetime
+
+        changes = []
+        for c in getattr(event, "changes", []):
+            changes.append(
+                SchemaChangeResult(
+                    change_type=c.change_type.value if hasattr(c.change_type, "value") else str(c.change_type),
+                    column_name=getattr(c, "column", getattr(c, "column_name", "")),
+                    old_value=getattr(c, "old_value", None),
+                    new_value=getattr(c, "new_value", None),
+                    severity=c.severity.value if hasattr(c.severity, "value") else str(c.severity),
+                    breaking=getattr(c, "breaking", False),
+                    description=getattr(c, "description", ""),
+                    migration_hint=getattr(c, "migration_hint", None),
+                )
+            )
+
+        timestamp = datetime.utcnow().isoformat()
+        if hasattr(event, "timestamp"):
+            timestamp = (
+                event.timestamp.isoformat()
+                if isinstance(event.timestamp, datetime)
+                else str(event.timestamp)
+            )
+
+        return SchemaWatcherEvent(
+            source=getattr(event, "source", ""),
+            has_breaking_changes=event.has_breaking_changes() if callable(getattr(event, "has_breaking_changes", None)) else getattr(event, "has_breaking_changes", False),
+            total_changes=len(changes),
+            changes=changes,
+            timestamp=timestamp,
+        )
+
+    def _convert_alert_result(self, alert: Any) -> BreakingChangeAlert:
+        """Convert truthound alert result."""
+        from datetime import datetime
+
+        changes = []
+        for c in getattr(alert, "changes", []):
+            if isinstance(c, dict):
+                changes.append(
+                    SchemaChangeResult(
+                        change_type=c.get("change_type", "unknown"),
+                        column_name=c.get("column_name", c.get("column", "")),
+                        old_value=c.get("old_value"),
+                        new_value=c.get("new_value"),
+                        severity=c.get("severity", "info"),
+                        breaking=c.get("breaking", False),
+                        description=c.get("description", ""),
+                        migration_hint=c.get("migration_hint"),
+                    )
+                )
+            else:
+                changes.append(
+                    SchemaChangeResult(
+                        change_type=c.change_type.value if hasattr(c.change_type, "value") else str(c.change_type),
+                        column_name=getattr(c, "column", getattr(c, "column_name", "")),
+                        old_value=getattr(c, "old_value", None),
+                        new_value=getattr(c, "new_value", None),
+                        severity=c.severity.value if hasattr(c.severity, "value") else str(c.severity),
+                        breaking=getattr(c, "breaking", False),
+                        description=getattr(c, "description", ""),
+                        migration_hint=getattr(c, "migration_hint", None),
+                    )
+                )
+
+        # Extract impact info
+        impact = getattr(alert, "impact", None)
+        impact_scope = "local"
+        affected_consumers: list[str] = []
+        data_risk_level = 1
+        recommendations: list[str] = []
+
+        if impact:
+            impact_scope = impact.scope.value if hasattr(impact.scope, "value") else str(impact.scope)
+            affected_consumers = list(getattr(impact, "affected_consumers", []))
+            data_risk_level = getattr(impact, "data_risk_level", 1)
+            recommendations = list(getattr(impact, "recommendations", []))
+
+        # Extract timestamps
+        def _format_dt(dt: Any) -> str | None:
+            if dt is None:
+                return None
+            if isinstance(dt, datetime):
+                return dt.isoformat()
+            return str(dt)
+
+        return BreakingChangeAlert(
+            alert_id=getattr(alert, "alert_id", ""),
+            title=getattr(alert, "title", ""),
+            source=getattr(alert, "source", ""),
+            changes=changes,
+            impact_scope=impact_scope,
+            affected_consumers=affected_consumers,
+            data_risk_level=data_risk_level,
+            recommendations=recommendations,
+            status=getattr(alert, "status", "open"),
+            created_at=_format_dt(getattr(alert, "created_at", None)) or datetime.utcnow().isoformat(),
+            acknowledged_at=_format_dt(getattr(alert, "acknowledged_at", None)),
+            resolved_at=_format_dt(getattr(alert, "resolved_at", None)),
+        )
+
+    def shutdown(self) -> None:
+        """Shutdown the executor and stop all watchers."""
+        # Stop all watchers
+        for watcher_id in list(self._watchers.keys()):
+            watcher_data = self._watchers[watcher_id]
+            if watcher_data["status"] == "running":
+                watcher_data["watcher"].stop()
+
+        self._watchers.clear()
+        self._histories.clear()
+        self._executor.shutdown(wait=False)
+
+
+# Singleton instance for schema evolution
+_schema_evolution_adapter: SchemaEvolutionAdapter | None = None
+
+
+def get_schema_evolution_adapter() -> SchemaEvolutionAdapter:
+    """Get singleton schema evolution adapter instance.
+
+    Returns:
+        SchemaEvolutionAdapter singleton.
+    """
+    global _schema_evolution_adapter
+    if _schema_evolution_adapter is None:
+        from truthound_dashboard.config import get_settings
+
+        settings = get_settings()
+        _schema_evolution_adapter = SchemaEvolutionAdapter(
+            max_workers=settings.max_workers
+        )
+    return _schema_evolution_adapter
+
+
+def reset_schema_evolution_adapter() -> None:
+    """Reset schema evolution adapter singleton (for testing)."""
+    global _schema_evolution_adapter
+    if _schema_evolution_adapter is not None:
+        _schema_evolution_adapter.shutdown()
+        _schema_evolution_adapter = None