tapps-agents 3.6.0__py3-none-any.whl → 3.6.1__py3-none-any.whl

tapps_agents/core/brownfield_review.py

@@ -1,541 +1,541 @@
-"""
-Brownfield System Review Orchestrator
-
-Orchestrates complete brownfield review workflow: analysis, expert creation, and RAG population.
-"""
-
-from __future__ import annotations
-
-import json
-import logging
-import time
-from dataclasses import dataclass, field
-from datetime import datetime
-from pathlib import Path
-from typing import Any
-
-from ..context7.agent_integration import Context7AgentHelper
-from ..experts.domain_detector import DomainMapping
-from ..experts.knowledge_ingestion import IngestionResult, KnowledgeIngestionPipeline
-from .brownfield_analyzer import BrownfieldAnalysisResult, BrownfieldAnalyzer
-from .expert_config_generator import ExpertConfig, ExpertConfigGenerator
-
-logger = logging.getLogger(__name__)
-
-
-@dataclass
-class BrownfieldReviewResult:
-    """Result of complete brownfield review."""
-
-    analysis: BrownfieldAnalysisResult
-    experts_created: list[ExpertConfig] = field(default_factory=list)
-    rag_results: dict[str, IngestionResult] = field(
-        default_factory=dict
-    )  # expert_id -> ingestion result
-    errors: list[str] = field(default_factory=list)
-    warnings: list[str] = field(default_factory=list)
-    execution_time: float = 0.0
-    dry_run: bool = False
-    report: str = ""
-
-
-class BrownfieldReviewOrchestrator:
-    """
-    Orchestrates complete brownfield review workflow.
-
-    Usage:
-        orchestrator = BrownfieldReviewOrchestrator(
-            project_root=Path.cwd(),
-            context7_helper=context7_helper,
-            dry_run=False
-        )
-        result = await orchestrator.review(auto=True, include_context7=True)
-        print(result.report)
-    """
-
-    def __init__(
-        self,
-        project_root: Path,
-        context7_helper: Context7AgentHelper | None = None,
-        dry_run: bool = False,
-    ) -> None:
-        """
-        Initialize orchestrator.
-
-        Args:
-            project_root: Root directory of project to review
-            context7_helper: Optional Context7 helper for library docs
-            dry_run: If True, preview changes without applying
-        """
-        self.project_root = Path(project_root).resolve()
-        self.context7_helper = context7_helper
-        self.dry_run = dry_run
-        self.config_dir = self.project_root / ".tapps-agents"
-        self.state_file = self.config_dir / "brownfield-review-state.json"
-
-        # Initialize components
-        self.analyzer = BrownfieldAnalyzer(project_root=self.project_root)
-        self.config_generator = ExpertConfigGenerator(project_root=self.project_root)
-        self.ingestion_pipeline = KnowledgeIngestionPipeline(
-            project_root=self.project_root, context7_helper=context7_helper
-        )
-
-    async def review(
-        self,
-        auto: bool = False,
-        include_context7: bool = True,
-        resume: bool = False,
-        resume_from: str | None = None,
-    ) -> BrownfieldReviewResult:
-        """
-        Perform complete brownfield review.
-
-        Args:
-            auto: If True, skip all prompts and use defaults
-            include_context7: If True, fetch library docs from Context7
-            resume: If True, resume from saved state
-            resume_from: Step to resume from ("analyze", "create_experts", "populate_rag")
-
-        Returns:
-            BrownfieldReviewResult with complete analysis and results
-        """
-        start_time = time.time()
-        errors: list[str] = []
-        warnings: list[str] = []
-
-        logger.info("Starting brownfield review")
-        if self.dry_run:
-            logger.info("DRY RUN MODE: No changes will be applied")
-
-        # Load state if resuming
-        state = None
-        if resume:
-            state = self._load_state()
-            if state and resume_from:
-                logger.info(f"Resuming from step: {resume_from}")
-
-        try:
-            # Step 1: Analyze codebase
-            if not resume or not state or resume_from == "analyze" or resume_from is None:
-                logger.info("Step 1: Analyzing codebase...")
-                analysis = await self._analyze_codebase()
-                self._save_state({"step": "analyze", "analysis": self._serialize_analysis(analysis)})
-                logger.info(
-                    f"Analysis complete: {len(analysis.domains)} domains detected"
-                )
-            else:
-                # Load from state
-                analysis = self._deserialize_analysis(state.get("analysis", {}))
-                logger.info("Loaded analysis from saved state")
-
-            # Step 2: Create expert configurations
-            if not resume or not state or resume_from == "create_experts" or resume_from is None:
-                logger.info("Step 2: Creating expert configurations...")
-                experts_created = await self._create_experts(analysis.domains)
-                self._save_state({
-                    "step": "create_experts",
-                    "analysis": self._serialize_analysis(analysis),
-                    "experts_created": [self._serialize_expert(e) for e in experts_created],
-                })
-                logger.info(f"Created {len(experts_created)} expert configurations")
-            else:
-                # Load from state
-                experts_created = [
-                    self._deserialize_expert(e) for e in state.get("experts_created", [])
-                ]
-                logger.info(f"Loaded {len(experts_created)} experts from saved state")
-
-            # Step 3: Populate RAG knowledge bases
-            if not resume or not state or resume_from == "populate_rag" or resume_from is None:
-                logger.info("Step 3: Populating RAG knowledge bases...")
-                rag_results = await self._populate_rag(
-                    experts_created, include_context7=include_context7
-                )
-                self._save_state({
-                    "step": "complete",
-                    "analysis": self._serialize_analysis(analysis),
-                    "experts_created": [self._serialize_expert(e) for e in experts_created],
-                    "rag_results": {k: {
-                        "entries_ingested": v.entries_ingested,
-                        "entries_failed": v.entries_failed,
-                        "source_type": v.source_type,
-                    } for k, v in rag_results.items()},
-                })
-                logger.info(
-                    f"Populated RAG for {len(rag_results)} experts"
-                )
-            else:
-                # Load from state
-                rag_results = {
-                    k: IngestionResult(
-                        source_type=v.get("source_type", "unknown"),
-                        entries_ingested=v.get("entries_ingested", 0),
-                        entries_failed=v.get("entries_failed", 0),
-                    )
-                    for k, v in state.get("rag_results", {}).items()
-                }
-                logger.info(f"Loaded RAG results from saved state")
-
-        except Exception as e:
-            error_msg = f"Brownfield review failed: {e}"
-            logger.error(error_msg, exc_info=True)
-            errors.append(error_msg)
-            # Save error state
-            self._save_state({
-                "step": "error",
-                "error": str(e),
-                "errors": errors,
-            })
-            # Create minimal result for error case
-            if 'analysis' not in locals():
-                analysis = BrownfieldAnalysisResult(project_root=self.project_root)
-            if 'experts_created' not in locals():
-                experts_created = []
-            if 'rag_results' not in locals():
-                rag_results = {}
-
-        execution_time = time.time() - start_time
-
-        # Generate report
-        result = BrownfieldReviewResult(
-            analysis=analysis,
-            experts_created=experts_created,
-            rag_results=rag_results,
-            errors=errors,
-            warnings=warnings,
-            execution_time=execution_time,
-            dry_run=self.dry_run,
-        )
-
-        result.report = self._generate_report(result)
-
-        logger.info(f"Brownfield review complete in {execution_time:.2f}s")
-        return result
-
-    async def _analyze_codebase(self) -> BrownfieldAnalysisResult:
-        """Step 1: Analyze codebase structure and detect domains."""
-        try:
-            result = await self.analyzer.analyze()
-            return result
-        except Exception as e:
-            logger.error(f"Codebase analysis failed: {e}", exc_info=True)
-            raise
-
-    async def _create_experts(
-        self, domains: list[DomainMapping]
-    ) -> list[ExpertConfig]:
-        """Step 2: Create expert configurations."""
-        if self.dry_run:
-            logger.info("DRY RUN: Would create expert configs")
-            # Still generate configs for preview
-            configs = self.config_generator.generate_expert_configs(domains)
-            return configs
-
-        try:
-            # Generate expert configurations
-            configs = self.config_generator.generate_expert_configs(domains)
-
-            # Validate each config
-            valid_configs = []
-            for config in configs:
-                if self.config_generator.validate_config(config):
-                    valid_configs.append(config)
-                else:
-                    logger.warning(f"Invalid config for {config.expert_id}, skipping")
-
-            # Write to YAML file
-            if valid_configs:
-                self.config_generator.write_expert_configs(valid_configs, merge=True)
-
-                # Create knowledge base directories
-                for config in valid_configs:
-                    if config.knowledge_base_dir:
-                        config.knowledge_base_dir.mkdir(parents=True, exist_ok=True)
-                        logger.info(
-                            f"Created knowledge base directory: {config.knowledge_base_dir}"
-                        )
-
-            return valid_configs
-        except Exception as e:
-            logger.error(f"Expert creation failed: {e}", exc_info=True)
-            raise
-
-    async def _populate_rag(
-        self, experts: list[ExpertConfig], include_context7: bool = True
-    ) -> dict[str, IngestionResult]:
-        """Step 3: Populate RAG knowledge bases for each expert."""
-        results: dict[str, IngestionResult] = {}
-
-        if self.dry_run:
-            logger.info("DRY RUN: Would populate RAG knowledge bases")
-            return results
-
-        # Use parallel processing for better performance
-        import asyncio
-
-        async def populate_expert_rag(expert: ExpertConfig) -> tuple[str, IngestionResult]:
-            """Populate RAG for a single expert."""
-            try:
-                logger.info(f"Populating RAG for expert {expert.expert_id}...")
-
-                # Create expert-specific knowledge base directory if needed
-                if expert.knowledge_base_dir:
-                    expert.knowledge_base_dir.mkdir(parents=True, exist_ok=True)
-
-                # Create expert-specific ingestion pipeline
-                expert_pipeline = KnowledgeIngestionPipeline(
-                    project_root=self.project_root,
-                    context7_helper=self.context7_helper,
-                )
-
-                # Ingest project sources for this expert's domain
-                project_result = expert_pipeline.ingest_project_sources()
-
-                # Store ingested content in expert-specific KB directory
-                if expert.knowledge_base_dir and project_result.entries_ingested > 0:
-                    # Copy/move ingested files to expert KB directory
-                    # This is a simplified version - full implementation would
-                    # use the knowledge base's add_chunk methods
-                    logger.debug(
-                        f"Stored {project_result.entries_ingested} entries in {expert.knowledge_base_dir}"
-                    )
-
-                # Ingest Context7 sources if enabled
-                context7_result = None
-                if include_context7 and self.context7_helper:
-                    try:
-                        context7_result = (
-                            await expert_pipeline.ingest_context7_sources()
-                        )
-                        # Combine project and context7 results
-                        if context7_result:
-                            project_result.entries_ingested += context7_result.entries_ingested
-                            project_result.entries_failed += context7_result.entries_failed
-                            project_result.errors.extend(context7_result.errors)
-                    except Exception as e:
-                        logger.warning(
-                            f"Context7 ingestion failed for {expert.expert_id}: {e}"
-                        )
-
-                logger.info(
-                    f"RAG population complete for {expert.expert_id}: "
-                    f"{project_result.entries_ingested} entries ingested"
-                )
-
-                return (expert.expert_id, project_result)
-
-            except Exception as e:
-                logger.error(
-                    f"RAG population failed for {expert.expert_id}: {e}",
-                    exc_info=True,
-                )
-                return (
-                    expert.expert_id,
-                    IngestionResult(
-                        source_type="error",
-                        entries_ingested=0,
-                        entries_failed=1,
-                        errors=[str(e)],
-                    ),
-                )
-
-        # Process experts in parallel (with limit to avoid overwhelming system)
-        tasks = [populate_expert_rag(expert) for expert in experts]
-        expert_results = await asyncio.gather(*tasks, return_exceptions=True)
-
-        # Collect results
-        for result in expert_results:
-            if isinstance(result, Exception):
-                logger.error(f"Expert RAG population task failed: {result}")
-                continue
-            expert_id, ingestion_result = result
-            results[expert_id] = ingestion_result
-
-        return results
-
-    def _generate_report(self, result: BrownfieldReviewResult) -> str:
-        """Generate human-readable summary report."""
-        lines = []
-        lines.append("# Brownfield System Review Report")
-        lines.append("")
-        lines.append(f"**Generated:** {datetime.now().isoformat()}")
-        lines.append(f"**Execution Time:** {result.execution_time:.2f}s")
-        if result.dry_run:
-            lines.append("**Mode:** DRY RUN (no changes applied)")
-        lines.append("")
-
-        # Analysis Summary
-        lines.append("## Analysis Summary")
-        lines.append("")
-        lines.append(f"- **Languages Detected:** {', '.join(result.analysis.languages) or 'None'}")
-        lines.append(f"- **Frameworks Detected:** {', '.join(result.analysis.frameworks) or 'None'}")
-        lines.append(f"- **Dependencies Found:** {len(result.analysis.dependencies)}")
-        lines.append(f"- **Domains Detected:** {len(result.analysis.domains)}")
-        lines.append("")
-
-        # Domains
-        if result.analysis.domains:
-            lines.append("### Detected Domains")
-            lines.append("")
-            for domain in result.analysis.domains[:10]:  # Top 10
-                lines.append(
-                    f"- **{domain.domain}** (confidence: {domain.confidence:.2f})"
-                )
-            if len(result.analysis.domains) > 10:
-                lines.append(f"- ... and {len(result.analysis.domains) - 10} more")
-            lines.append("")
-
-        # Experts Created
-        lines.append("## Experts Created")
-        lines.append("")
-        if result.experts_created:
-            for expert in result.experts_created:
-                lines.append(f"- **{expert.expert_name}** (`{expert.expert_id}`)")
-                lines.append(f"  - Domain: {expert.primary_domain}")
-                lines.append(f"  - RAG Enabled: {expert.rag_enabled}")
-                if expert.knowledge_base_dir:
-                    lines.append(f"  - KB Directory: `{expert.knowledge_base_dir}`")
-            lines.append("")
-        else:
-            lines.append("No new experts created (may already exist or no domains detected)")
-            lines.append("")
-
-        # RAG Population Results
-        lines.append("## RAG Population Results")
-        lines.append("")
-        if result.rag_results:
-            total_ingested = sum(
-                r.entries_ingested for r in result.rag_results.values()
-            )
-            total_failed = sum(r.entries_failed for r in result.rag_results.values())
-            lines.append(f"- **Total Entries Ingested:** {total_ingested}")
-            lines.append(f"- **Total Entries Failed:** {total_failed}")
-            lines.append("")
-            for expert_id, rag_result in result.rag_results.items():
-                lines.append(f"### {expert_id}")
-                lines.append(f"- Ingested: {rag_result.entries_ingested}")
-                lines.append(f"- Failed: {rag_result.entries_failed}")
-                if rag_result.errors:
-                    lines.append(f"- Errors: {len(rag_result.errors)}")
-                lines.append("")
-        else:
-            lines.append("No RAG population performed (dry run or no experts)")
-            lines.append("")
-
-        # Errors and Warnings
-        if result.errors:
-            lines.append("## Errors")
-            lines.append("")
-            for error in result.errors:
-                lines.append(f"- ❌ {error}")
-            lines.append("")
-
-        if result.warnings:
-            lines.append("## Warnings")
-            lines.append("")
-            for warning in result.warnings:
-                lines.append(f"- ⚠️ {warning}")
-            lines.append("")
-
-        # Summary
-        lines.append("## Summary")
-        lines.append("")
-        if result.dry_run:
-            lines.append("This was a dry run. No changes were applied.")
-        else:
-            lines.append(
-                f"✅ Created {len(result.experts_created)} expert configurations"
-            )
-            lines.append(
-                f"✅ Populated RAG for {len(result.rag_results)} experts"
-            )
-        lines.append("")
-
-        return "\n".join(lines)
-
-    def _save_state(self, state: dict[str, Any]) -> None:
-        """Save workflow state to file."""
-        if self.dry_run:
-            return  # Don't save state in dry-run mode
-
-        try:
-            self.config_dir.mkdir(parents=True, exist_ok=True)
-            with open(self.state_file, "w", encoding="utf-8") as f:
-                json.dump(state, f, indent=2, default=str)
-            logger.debug(f"Saved state to {self.state_file}")
-        except Exception as e:
-            logger.warning(f"Failed to save state: {e}")
-
-    def _load_state(self) -> dict[str, Any] | None:
-        """Load workflow state from file."""
-        if not self.state_file.exists():
-            return None
-
-        try:
-            with open(self.state_file, "r", encoding="utf-8") as f:
-                return json.load(f)
-        except Exception as e:
-            logger.warning(f"Failed to load state: {e}")
-            return None
-
-    def _serialize_analysis(self, analysis: BrownfieldAnalysisResult) -> dict[str, Any]:
-        """Serialize analysis result for state storage."""
-        return {
-            "languages": analysis.languages,
-            "frameworks": analysis.frameworks,
-            "dependencies": analysis.dependencies,
-            "domains": [
-                {
-                    "domain": d.domain,
-                    "confidence": d.confidence,
-                    "reasoning": d.reasoning,
-                }
-                for d in analysis.domains
-            ],
-            "analysis_metadata": analysis.analysis_metadata,
-        }
-
-    def _deserialize_analysis(self, data: dict[str, Any]) -> BrownfieldAnalysisResult:
-        """Deserialize analysis result from state."""
-        from ..experts.domain_detector import DomainMapping
-
-        domains = [
-            DomainMapping(
-                domain=d["domain"],
-                confidence=d["confidence"],
-                signals=[],
-                reasoning=d.get("reasoning", ""),
-            )
-            for d in data.get("domains", [])
-        ]
-
-        return BrownfieldAnalysisResult(
-            project_root=self.project_root,
-            languages=data.get("languages", []),
-            frameworks=data.get("frameworks", []),
-            dependencies=data.get("dependencies", []),
-            domains=domains,
-            analysis_metadata=data.get("analysis_metadata", {}),
-        )
-
-    def _serialize_expert(self, expert: ExpertConfig) -> dict[str, Any]:
-        """Serialize expert config for state storage."""
-        return {
-            "expert_id": expert.expert_id,
-            "expert_name": expert.expert_name,
-            "primary_domain": expert.primary_domain,
-            "rag_enabled": expert.rag_enabled,
-            "knowledge_base_dir": str(expert.knowledge_base_dir) if expert.knowledge_base_dir else None,
-            "confidence_matrix": expert.confidence_matrix,
-        }
-
-    def _deserialize_expert(self, data: dict[str, Any]) -> ExpertConfig:
-        """Deserialize expert config from state."""
-        return ExpertConfig(
-            expert_id=data["expert_id"],
-            expert_name=data["expert_name"],
-            primary_domain=data["primary_domain"],
-            rag_enabled=data.get("rag_enabled", True),
-            knowledge_base_dir=Path(data["knowledge_base_dir"]) if data.get("knowledge_base_dir") else None,
-            confidence_matrix=data.get("confidence_matrix"),
-        )
+"""
+Brownfield System Review Orchestrator
+
+Orchestrates complete brownfield review workflow: analysis, expert creation, and RAG population.
+"""
+
+from __future__ import annotations
+
+import json
+import logging
+import time
+from dataclasses import dataclass, field
+from datetime import datetime
+from pathlib import Path
+from typing import Any
+
+from ..context7.agent_integration import Context7AgentHelper
+from ..experts.domain_detector import DomainMapping
+from ..experts.knowledge_ingestion import IngestionResult, KnowledgeIngestionPipeline
+from .brownfield_analyzer import BrownfieldAnalysisResult, BrownfieldAnalyzer
+from .expert_config_generator import ExpertConfig, ExpertConfigGenerator
+
+logger = logging.getLogger(__name__)
+
+
+@dataclass
+class BrownfieldReviewResult:
+    """Result of complete brownfield review."""
+
+    analysis: BrownfieldAnalysisResult
+    experts_created: list[ExpertConfig] = field(default_factory=list)
+    rag_results: dict[str, IngestionResult] = field(
+        default_factory=dict
+    )  # expert_id -> ingestion result
+    errors: list[str] = field(default_factory=list)
+    warnings: list[str] = field(default_factory=list)
+    execution_time: float = 0.0
+    dry_run: bool = False
+    report: str = ""
+
+
+class BrownfieldReviewOrchestrator:
+    """
+    Orchestrates complete brownfield review workflow.
+
+    Usage:
+        orchestrator = BrownfieldReviewOrchestrator(
+            project_root=Path.cwd(),
+            context7_helper=context7_helper,
+            dry_run=False
+        )
+        result = await orchestrator.review(auto=True, include_context7=True)
+        print(result.report)
+    """
+
+    def __init__(
+        self,
+        project_root: Path,
+        context7_helper: Context7AgentHelper | None = None,
+        dry_run: bool = False,
+    ) -> None:
+        """
+        Initialize orchestrator.
+
+        Args:
+            project_root: Root directory of project to review
+            context7_helper: Optional Context7 helper for library docs
+            dry_run: If True, preview changes without applying
+        """
+        self.project_root = Path(project_root).resolve()
+        self.context7_helper = context7_helper
+        self.dry_run = dry_run
+        self.config_dir = self.project_root / ".tapps-agents"
+        self.state_file = self.config_dir / "brownfield-review-state.json"
+
+        # Initialize components
+        self.analyzer = BrownfieldAnalyzer(project_root=self.project_root)
+        self.config_generator = ExpertConfigGenerator(project_root=self.project_root)
+        self.ingestion_pipeline = KnowledgeIngestionPipeline(
+            project_root=self.project_root, context7_helper=context7_helper
+        )
+
+    async def review(
+        self,
+        auto: bool = False,
+        include_context7: bool = True,
+        resume: bool = False,
+        resume_from: str | None = None,
+    ) -> BrownfieldReviewResult:
+        """
+        Perform complete brownfield review.
+
+        Args:
+            auto: If True, skip all prompts and use defaults
+            include_context7: If True, fetch library docs from Context7
+            resume: If True, resume from saved state
+            resume_from: Step to resume from ("analyze", "create_experts", "populate_rag")
+
+        Returns:
+            BrownfieldReviewResult with complete analysis and results
+        """
+        start_time = time.time()
+        errors: list[str] = []
+        warnings: list[str] = []
+
+        logger.info("Starting brownfield review")
+        if self.dry_run:
+            logger.info("DRY RUN MODE: No changes will be applied")
+
+        # Load state if resuming
+        state = None
+        if resume:
+            state = self._load_state()
+            if state and resume_from:
+                logger.info(f"Resuming from step: {resume_from}")
+
+        try:
+            # Step 1: Analyze codebase
+            if not resume or not state or resume_from == "analyze" or resume_from is None:
+                logger.info("Step 1: Analyzing codebase...")
+                analysis = await self._analyze_codebase()
+                self._save_state({"step": "analyze", "analysis": self._serialize_analysis(analysis)})
+                logger.info(
+                    f"Analysis complete: {len(analysis.domains)} domains detected"
+                )
+            else:
+                # Load from state
+                analysis = self._deserialize_analysis(state.get("analysis", {}))
+                logger.info("Loaded analysis from saved state")
+
+            # Step 2: Create expert configurations
+            if not resume or not state or resume_from == "create_experts" or resume_from is None:
+                logger.info("Step 2: Creating expert configurations...")
+                experts_created = await self._create_experts(analysis.domains)
+                self._save_state({
+                    "step": "create_experts",
+                    "analysis": self._serialize_analysis(analysis),
+                    "experts_created": [self._serialize_expert(e) for e in experts_created],
+                })
+                logger.info(f"Created {len(experts_created)} expert configurations")
+            else:
+                # Load from state
+                experts_created = [
+                    self._deserialize_expert(e) for e in state.get("experts_created", [])
+                ]
+                logger.info(f"Loaded {len(experts_created)} experts from saved state")
+
+            # Step 3: Populate RAG knowledge bases
+            if not resume or not state or resume_from == "populate_rag" or resume_from is None:
+                logger.info("Step 3: Populating RAG knowledge bases...")
+                rag_results = await self._populate_rag(
+                    experts_created, include_context7=include_context7
+                )
+                self._save_state({
+                    "step": "complete",
+                    "analysis": self._serialize_analysis(analysis),
+                    "experts_created": [self._serialize_expert(e) for e in experts_created],
+                    "rag_results": {k: {
+                        "entries_ingested": v.entries_ingested,
+                        "entries_failed": v.entries_failed,
+                        "source_type": v.source_type,
+                    } for k, v in rag_results.items()},
+                })
+                logger.info(
+                    f"Populated RAG for {len(rag_results)} experts"
+                )
+            else:
+                # Load from state
+                rag_results = {
+                    k: IngestionResult(
+                        source_type=v.get("source_type", "unknown"),
+                        entries_ingested=v.get("entries_ingested", 0),
+                        entries_failed=v.get("entries_failed", 0),
+                    )
+                    for k, v in state.get("rag_results", {}).items()
+                }
+                logger.info(f"Loaded RAG results from saved state")
+
+        except Exception as e:
+            error_msg = f"Brownfield review failed: {e}"
+            logger.error(error_msg, exc_info=True)
+            errors.append(error_msg)
+            # Save error state
+            self._save_state({
+                "step": "error",
+                "error": str(e),
+                "errors": errors,
+            })
+            # Create minimal result for error case
+            if 'analysis' not in locals():
+                analysis = BrownfieldAnalysisResult(project_root=self.project_root)
+            if 'experts_created' not in locals():
+                experts_created = []
+            if 'rag_results' not in locals():
+                rag_results = {}
+
+        execution_time = time.time() - start_time
+
+        # Generate report
+        result = BrownfieldReviewResult(
+            analysis=analysis,
+            experts_created=experts_created,
+            rag_results=rag_results,
+            errors=errors,
+            warnings=warnings,
+            execution_time=execution_time,
+            dry_run=self.dry_run,
+        )
+
+        result.report = self._generate_report(result)
+
+        logger.info(f"Brownfield review complete in {execution_time:.2f}s")
+        return result
+
+    async def _analyze_codebase(self) -> BrownfieldAnalysisResult:
+        """Step 1: Analyze codebase structure and detect domains."""
+        try:
+            result = await self.analyzer.analyze()
+            return result
+        except Exception as e:
+            logger.error(f"Codebase analysis failed: {e}", exc_info=True)
+            raise
+
+    async def _create_experts(
+        self, domains: list[DomainMapping]
+    ) -> list[ExpertConfig]:
+        """Step 2: Create expert configurations."""
+        if self.dry_run:
+            logger.info("DRY RUN: Would create expert configs")
+            # Still generate configs for preview
+            configs = self.config_generator.generate_expert_configs(domains)
+            return configs
+
+        try:
+            # Generate expert configurations
+            configs = self.config_generator.generate_expert_configs(domains)
+
+            # Validate each config
+            valid_configs = []
+            for config in configs:
+                if self.config_generator.validate_config(config):
+                    valid_configs.append(config)
+                else:
+                    logger.warning(f"Invalid config for {config.expert_id}, skipping")
+
+            # Write to YAML file
+            if valid_configs:
+                self.config_generator.write_expert_configs(valid_configs, merge=True)
+
+                # Create knowledge base directories
+                for config in valid_configs:
+                    if config.knowledge_base_dir:
+                        config.knowledge_base_dir.mkdir(parents=True, exist_ok=True)
+                        logger.info(
+                            f"Created knowledge base directory: {config.knowledge_base_dir}"
+                        )
+
+            return valid_configs
+        except Exception as e:
+            logger.error(f"Expert creation failed: {e}", exc_info=True)
+            raise
+
+    async def _populate_rag(
+        self, experts: list[ExpertConfig], include_context7: bool = True
+    ) -> dict[str, IngestionResult]:
+        """Step 3: Populate RAG knowledge bases for each expert."""
+        results: dict[str, IngestionResult] = {}
+
+        if self.dry_run:
+            logger.info("DRY RUN: Would populate RAG knowledge bases")
+            return results
+
+        # Use parallel processing for better performance
+        import asyncio
+
+        async def populate_expert_rag(expert: ExpertConfig) -> tuple[str, IngestionResult]:
+            """Populate RAG for a single expert."""
+            try:
+                logger.info(f"Populating RAG for expert {expert.expert_id}...")
+
+                # Create expert-specific knowledge base directory if needed
+                if expert.knowledge_base_dir:
+                    expert.knowledge_base_dir.mkdir(parents=True, exist_ok=True)
+
+                # Create expert-specific ingestion pipeline
+                expert_pipeline = KnowledgeIngestionPipeline(
+                    project_root=self.project_root,
+                    context7_helper=self.context7_helper,
+                )
+
+                # Ingest project sources for this expert's domain
+                project_result = expert_pipeline.ingest_project_sources()
+
+                # Store ingested content in expert-specific KB directory
+                if expert.knowledge_base_dir and project_result.entries_ingested > 0:
+                    # Copy/move ingested files to expert KB directory
+                    # This is a simplified version - full implementation would
+                    # use the knowledge base's add_chunk methods
+                    logger.debug(
+                        f"Stored {project_result.entries_ingested} entries in {expert.knowledge_base_dir}"
+                    )
+
+                # Ingest Context7 sources if enabled
+                context7_result = None
+                if include_context7 and self.context7_helper:
+                    try:
+                        context7_result = (
+                            await expert_pipeline.ingest_context7_sources()
+                        )
+                        # Combine project and context7 results
+                        if context7_result:
+                            project_result.entries_ingested += context7_result.entries_ingested
+                            project_result.entries_failed += context7_result.entries_failed
+                            project_result.errors.extend(context7_result.errors)
+                    except Exception as e:
+                        logger.warning(
+                            f"Context7 ingestion failed for {expert.expert_id}: {e}"
+                        )
+
+                logger.info(
+                    f"RAG population complete for {expert.expert_id}: "
+                    f"{project_result.entries_ingested} entries ingested"
+                )
+
+                return (expert.expert_id, project_result)
+
+            except Exception as e:
+                logger.error(
+                    f"RAG population failed for {expert.expert_id}: {e}",
+                    exc_info=True,
+                )
+                return (
+                    expert.expert_id,
+                    IngestionResult(
+                        source_type="error",
+                        entries_ingested=0,
+                        entries_failed=1,
+                        errors=[str(e)],
+                    ),
+                )
+
+        # Process experts in parallel (with limit to avoid overwhelming system)
+        tasks = [populate_expert_rag(expert) for expert in experts]
+        expert_results = await asyncio.gather(*tasks, return_exceptions=True)
+
+        # Collect results
+        for result in expert_results:
+            if isinstance(result, Exception):
+                logger.error(f"Expert RAG population task failed: {result}")
+                continue
+            expert_id, ingestion_result = result
+            results[expert_id] = ingestion_result
+
+        return results
+
+    def _generate_report(self, result: BrownfieldReviewResult) -> str:
+        """Generate human-readable summary report."""
+        lines = []
+        lines.append("# Brownfield System Review Report")
+        lines.append("")
+        lines.append(f"**Generated:** {datetime.now().isoformat()}")
+        lines.append(f"**Execution Time:** {result.execution_time:.2f}s")
+        if result.dry_run:
+            lines.append("**Mode:** DRY RUN (no changes applied)")
+        lines.append("")
+
+        # Analysis Summary
+        lines.append("## Analysis Summary")
+        lines.append("")
+        lines.append(f"- **Languages Detected:** {', '.join(result.analysis.languages) or 'None'}")
+        lines.append(f"- **Frameworks Detected:** {', '.join(result.analysis.frameworks) or 'None'}")
+        lines.append(f"- **Dependencies Found:** {len(result.analysis.dependencies)}")
+        lines.append(f"- **Domains Detected:** {len(result.analysis.domains)}")
+        lines.append("")
+
+        # Domains
+        if result.analysis.domains:
+            lines.append("### Detected Domains")
+            lines.append("")
+            for domain in result.analysis.domains[:10]:  # Top 10
+                lines.append(
+                    f"- **{domain.domain}** (confidence: {domain.confidence:.2f})"
+                )
+            if len(result.analysis.domains) > 10:
+                lines.append(f"- ... and {len(result.analysis.domains) - 10} more")
+            lines.append("")
+
+        # Experts Created
+        lines.append("## Experts Created")
+        lines.append("")
+        if result.experts_created:
+            for expert in result.experts_created:
+                lines.append(f"- **{expert.expert_name}** (`{expert.expert_id}`)")
+                lines.append(f"  - Domain: {expert.primary_domain}")
+                lines.append(f"  - RAG Enabled: {expert.rag_enabled}")
+                if expert.knowledge_base_dir:
+                    lines.append(f"  - KB Directory: `{expert.knowledge_base_dir}`")
+            lines.append("")
+        else:
+            lines.append("No new experts created (may already exist or no domains detected)")
+            lines.append("")
+
+        # RAG Population Results
+        lines.append("## RAG Population Results")
+        lines.append("")
+        if result.rag_results:
+            total_ingested = sum(
+                r.entries_ingested for r in result.rag_results.values()
+            )
+            total_failed = sum(r.entries_failed for r in result.rag_results.values())
+            lines.append(f"- **Total Entries Ingested:** {total_ingested}")
+            lines.append(f"- **Total Entries Failed:** {total_failed}")
+            lines.append("")
+            for expert_id, rag_result in result.rag_results.items():
+                lines.append(f"### {expert_id}")
+                lines.append(f"- Ingested: {rag_result.entries_ingested}")
+                lines.append(f"- Failed: {rag_result.entries_failed}")
+                if rag_result.errors:
+                    lines.append(f"- Errors: {len(rag_result.errors)}")
+                lines.append("")
+        else:
+            lines.append("No RAG population performed (dry run or no experts)")
+            lines.append("")
+
+        # Errors and Warnings
+        if result.errors:
+            lines.append("## Errors")
+            lines.append("")
+            for error in result.errors:
+                lines.append(f"- ❌ {error}")
+            lines.append("")
+
+        if result.warnings:
+            lines.append("## Warnings")
+            lines.append("")
+            for warning in result.warnings:
+                lines.append(f"- ⚠️ {warning}")
+            lines.append("")
+
+        # Summary
+        lines.append("## Summary")
+        lines.append("")
+        if result.dry_run:
+            lines.append("This was a dry run. No changes were applied.")
+        else:
+            lines.append(
+                f"✅ Created {len(result.experts_created)} expert configurations"
+            )
+            lines.append(
+                f"✅ Populated RAG for {len(result.rag_results)} experts"
+            )
+        lines.append("")
+
+        return "\n".join(lines)
+
+    def _save_state(self, state: dict[str, Any]) -> None:
+        """Save workflow state to file."""
+        if self.dry_run:
+            return  # Don't save state in dry-run mode
+
+        try:
+            self.config_dir.mkdir(parents=True, exist_ok=True)
+            with open(self.state_file, "w", encoding="utf-8") as f:
+                json.dump(state, f, indent=2, default=str)
+            logger.debug(f"Saved state to {self.state_file}")
+        except Exception as e:
+            logger.warning(f"Failed to save state: {e}")
+
+    def _load_state(self) -> dict[str, Any] | None:
+        """Load workflow state from file."""
+        if not self.state_file.exists():
+            return None
+
+        try:
+            with open(self.state_file, "r", encoding="utf-8") as f:
+                return json.load(f)
+        except Exception as e:
+            logger.warning(f"Failed to load state: {e}")
+            return None
+
+    def _serialize_analysis(self, analysis: BrownfieldAnalysisResult) -> dict[str, Any]:
+        """Serialize analysis result for state storage."""
+        return {
+            "languages": analysis.languages,
+            "frameworks": analysis.frameworks,
+            "dependencies": analysis.dependencies,
+            "domains": [
+                {
+                    "domain": d.domain,
+                    "confidence": d.confidence,
+                    "reasoning": d.reasoning,
+                }
+                for d in analysis.domains
+            ],
+            "analysis_metadata": analysis.analysis_metadata,
+        }
+
+    def _deserialize_analysis(self, data: dict[str, Any]) -> BrownfieldAnalysisResult:
+        """Deserialize analysis result from state."""
+        from ..experts.domain_detector import DomainMapping
+
+        domains = [
+            DomainMapping(
+                domain=d["domain"],
+                confidence=d["confidence"],
+                signals=[],
+                reasoning=d.get("reasoning", ""),
+            )
+            for d in data.get("domains", [])
+        ]
+
+        return BrownfieldAnalysisResult(
+            project_root=self.project_root,
+            languages=data.get("languages", []),
+            frameworks=data.get("frameworks", []),
+            dependencies=data.get("dependencies", []),
+            domains=domains,
+            analysis_metadata=data.get("analysis_metadata", {}),
+        )
+
+    def _serialize_expert(self, expert: ExpertConfig) -> dict[str, Any]:
+        """Serialize expert config for state storage."""
+        return {
+            "expert_id": expert.expert_id,
+            "expert_name": expert.expert_name,
+            "primary_domain": expert.primary_domain,
+            "rag_enabled": expert.rag_enabled,
+            "knowledge_base_dir": str(expert.knowledge_base_dir) if expert.knowledge_base_dir else None,
+            "confidence_matrix": expert.confidence_matrix,
+        }
+
+    def _deserialize_expert(self, data: dict[str, Any]) -> ExpertConfig:
+        """Deserialize expert config from state."""
+        return ExpertConfig(
+            expert_id=data["expert_id"],
+            expert_name=data["expert_name"],
+            primary_domain=data["primary_domain"],
+            rag_enabled=data.get("rag_enabled", True),
+            knowledge_base_dir=Path(data["knowledge_base_dir"]) if data.get("knowledge_base_dir") else None,
+            confidence_matrix=data.get("confidence_matrix"),
+        )