kailash 0.8.1__py3-none-any.whl → 0.8.4__py3-none-any.whl

kailash/nodes/ai/a2a.py

@@ -13,13 +13,987 @@ import json
 import time
 import uuid
 from collections import defaultdict, deque
+from dataclasses import dataclass, field
 from datetime import datetime
-from typing import Any, Dict, List, Optional, Set
+from enum import Enum
+from typing import Any, Dict, List, Optional, Set, Tuple
 
 from kailash.nodes.ai.llm_agent import LLMAgentNode
 from kailash.nodes.base import Node, NodeParameter, register_node
 from kailash.nodes.base_cycle_aware import CycleAwareNode
 
+# ============================================================================
+# ENHANCED A2A COMPONENTS: Agent Cards and Task Management
+# ============================================================================
+
+
+class CapabilityLevel(Enum):
+    """Agent capability proficiency levels."""
+
+    NOVICE = "novice"
+    INTERMEDIATE = "intermediate"
+    ADVANCED = "advanced"
+    EXPERT = "expert"
+
+
+class CollaborationStyle(Enum):
+    """Agent collaboration preferences."""
+
+    INDEPENDENT = "independent"  # Prefers solo work
+    COOPERATIVE = "cooperative"  # Works well in teams
+    LEADER = "leader"  # Takes charge of coordination
+    SUPPORT = "support"  # Provides assistance to others
+
+
+class TaskState(Enum):
+    """Task lifecycle states."""
+
+    CREATED = "created"
+    ASSIGNED = "assigned"
+    IN_PROGRESS = "in_progress"
+    AWAITING_REVIEW = "awaiting_review"
+    ITERATING = "iterating"
+    COMPLETED = "completed"
+    FAILED = "failed"
+    CANCELLED = "cancelled"
+
+
+class TaskPriority(Enum):
+    """Task priority levels."""
+
+    LOW = "low"
+    MEDIUM = "medium"
+    HIGH = "high"
+    CRITICAL = "critical"
+
+
+class InsightType(Enum):
+    """Types of insights that can be generated."""
+
+    DISCOVERY = "discovery"  # New finding
+    ANALYSIS = "analysis"  # Deep analysis result
+    RECOMMENDATION = "recommendation"  # Actionable recommendation
+    WARNING = "warning"  # Potential issue
+    OPPORTUNITY = "opportunity"  # Improvement opportunity
+    PATTERN = "pattern"  # Identified pattern
+    ANOMALY = "anomaly"  # Unusual finding
+
+
+@dataclass
+class Capability:
+    """Detailed capability description."""
+
+    name: str
+    domain: str
+    level: CapabilityLevel
+    description: str
+    keywords: List[str] = field(default_factory=list)
+    examples: List[str] = field(default_factory=list)
+    constraints: List[str] = field(default_factory=list)
+
+    def matches_requirement(self, requirement: str) -> float:
+        """Calculate match score for a requirement (0.0-1.0)."""
+        requirement_lower = requirement.lower()
+
+        # Direct name match
+        if self.name.lower() in requirement_lower:
+            return 0.9
+
+        # Domain match
+        if self.domain.lower() in requirement_lower:
+            return 0.7
+
+        # Keyword matches
+        keyword_matches = sum(
+            1 for keyword in self.keywords if keyword.lower() in requirement_lower
+        )
+        if keyword_matches > 0:
+            return min(0.6 + (keyword_matches * 0.1), 0.8)
+
+        # Description similarity
+        desc_words = set(self.description.lower().split())
+        req_words = set(requirement_lower.split())
+        overlap = len(desc_words & req_words)
+        if overlap > 0:
+            return min(0.3 + (overlap * 0.05), 0.5)
+
+        return 0.0
+
+
+@dataclass
+class PerformanceMetrics:
+    """Agent performance tracking."""
+
+    total_tasks: int = 0
+    successful_tasks: int = 0
+    failed_tasks: int = 0
+
+    average_response_time_ms: float = 0.0
+    average_insight_quality: float = 0.0
+    average_confidence_score: float = 0.0
+
+    insights_generated: int = 0
+    unique_insights: int = 0
+    actionable_insights: int = 0
+
+    collaboration_score: float = 0.0
+    reliability_score: float = 0.0
+
+    last_active: Optional[datetime] = None
+
+    @property
+    def success_rate(self) -> float:
+        """Calculate task success rate."""
+        if self.total_tasks == 0:
+            return 0.0
+        return self.successful_tasks / self.total_tasks
+
+    @property
+    def insight_quality_score(self) -> float:
+        """Calculate overall insight quality score."""
+        if self.insights_generated == 0:
+            return 0.0
+
+        uniqueness = self.unique_insights / self.insights_generated
+        actionability = self.actionable_insights / self.insights_generated
+
+        return (
+            self.average_insight_quality * 0.4 + uniqueness * 0.3 + actionability * 0.3
+        )
+
+
+@dataclass
+class ResourceRequirements:
+    """Agent resource constraints and requirements."""
+
+    min_memory_mb: int = 512
+    max_memory_mb: int = 4096
+
+    min_tokens: int = 100
+    max_tokens: int = 4000
+
+    requires_gpu: bool = False
+    requires_internet: bool = True
+
+    estimated_cost_per_task: float = 0.0
+    max_concurrent_tasks: int = 5
+
+    supported_models: List[str] = field(default_factory=list)
+    required_apis: List[str] = field(default_factory=list)
+
+
+@dataclass
+class A2AAgentCard:
+    """
+    Enhanced agent card for rich capability description.
+
+    Provides comprehensive agent metadata for optimal matching,
+    team formation, and performance tracking.
+    """
+
+    # Identity
+    agent_id: str
+    agent_name: str
+    agent_type: str
+    version: str
+
+    # Capabilities
+    primary_capabilities: List[Capability] = field(default_factory=list)
+    secondary_capabilities: List[Capability] = field(default_factory=list)
+    emerging_capabilities: List[Capability] = field(default_factory=list)
+
+    # Collaboration
+    collaboration_style: CollaborationStyle = CollaborationStyle.COOPERATIVE
+    preferred_team_size: int = 3
+    compatible_agents: List[str] = field(default_factory=list)
+    incompatible_agents: List[str] = field(default_factory=list)
+
+    # Performance
+    performance: PerformanceMetrics = field(default_factory=PerformanceMetrics)
+
+    # Resources
+    resources: ResourceRequirements = field(default_factory=ResourceRequirements)
+
+    # Metadata
+    description: str = ""
+    tags: List[str] = field(default_factory=list)
+    created_at: datetime = field(default_factory=datetime.now)
+    updated_at: datetime = field(default_factory=datetime.now)
+
+    # Specializations
+    specializations: Dict[str, Any] = field(default_factory=dict)
+
+    def to_dict(self) -> Dict[str, Any]:
+        """Convert to dictionary for serialization."""
+        return {
+            "agent_id": self.agent_id,
+            "agent_name": self.agent_name,
+            "agent_type": self.agent_type,
+            "version": self.version,
+            "primary_capabilities": [
+                {
+                    "name": cap.name,
+                    "domain": cap.domain,
+                    "level": cap.level.value,
+                    "description": cap.description,
+                    "keywords": cap.keywords,
+                }
+                for cap in self.primary_capabilities
+            ],
+            "secondary_capabilities": [
+                {
+                    "name": cap.name,
+                    "domain": cap.domain,
+                    "level": cap.level.value,
+                    "description": cap.description,
+                }
+                for cap in self.secondary_capabilities
+            ],
+            "collaboration_style": self.collaboration_style.value,
+            "performance": {
+                "success_rate": self.performance.success_rate,
+                "insight_quality_score": self.performance.insight_quality_score,
+                "average_response_time_ms": self.performance.average_response_time_ms,
+                "reliability_score": self.performance.reliability_score,
+            },
+            "resources": {
+                "max_tokens": self.resources.max_tokens,
+                "requires_gpu": self.resources.requires_gpu,
+                "estimated_cost_per_task": self.resources.estimated_cost_per_task,
+            },
+            "tags": self.tags,
+            "description": self.description,
+        }
+
+    @classmethod
+    def from_dict(cls, data: Dict[str, Any]) -> "A2AAgentCard":
+        """Create from dictionary."""
+        # Convert capability dictionaries back to objects
+        primary_caps = [
+            Capability(
+                name=cap["name"],
+                domain=cap["domain"],
+                level=CapabilityLevel(cap["level"]),
+                description=cap.get("description", ""),
+                keywords=cap.get("keywords", []),
+            )
+            for cap in data.get("primary_capabilities", [])
+        ]
+
+        secondary_caps = [
+            Capability(
+                name=cap["name"],
+                domain=cap["domain"],
+                level=CapabilityLevel(cap["level"]),
+                description=cap.get("description", ""),
+            )
+            for cap in data.get("secondary_capabilities", [])
+        ]
+
+        # Create performance metrics
+        perf_data = data.get("performance", {})
+        performance = PerformanceMetrics()
+        if perf_data:
+            performance.average_insight_quality = perf_data.get(
+                "insight_quality_score", 0.0
+            )
+            performance.average_response_time_ms = perf_data.get(
+                "average_response_time_ms", 0.0
+            )
+            performance.reliability_score = perf_data.get("reliability_score", 0.0)
+
+        # Create resource requirements
+        res_data = data.get("resources", {})
+        resources = ResourceRequirements()
+        if res_data:
+            resources.max_tokens = res_data.get("max_tokens", 4000)
+            resources.requires_gpu = res_data.get("requires_gpu", False)
+            resources.estimated_cost_per_task = res_data.get(
+                "estimated_cost_per_task", 0.0
+            )
+
+        return cls(
+            agent_id=data["agent_id"],
+            agent_name=data["agent_name"],
+            agent_type=data["agent_type"],
+            version=data.get("version", "1.0.0"),
+            primary_capabilities=primary_caps,
+            secondary_capabilities=secondary_caps,
+            collaboration_style=CollaborationStyle(
+                data.get("collaboration_style", "cooperative")
+            ),
+            performance=performance,
+            resources=resources,
+            tags=data.get("tags", []),
+            description=data.get("description", ""),
+        )
+
+    def calculate_match_score(self, requirements: List[str]) -> float:
+        """
+        Calculate how well this agent matches given requirements.
+
+        Returns a score between 0.0 and 1.0.
+        """
+        if not requirements:
+            return 0.5  # Neutral score for no requirements
+
+        total_score = 0.0
+
+        for requirement in requirements:
+            # Check primary capabilities (highest weight)
+            primary_scores = [
+                cap.matches_requirement(requirement) * 1.0
+                for cap in self.primary_capabilities
+            ]
+
+            # Check secondary capabilities (medium weight)
+            secondary_scores = [
+                cap.matches_requirement(requirement) * 0.7
+                for cap in self.secondary_capabilities
+            ]
+
+            # Check emerging capabilities (lower weight)
+            emerging_scores = [
+                cap.matches_requirement(requirement) * 0.4
+                for cap in self.emerging_capabilities
+            ]
+
+            # Take the best match for this requirement
+            all_scores = primary_scores + secondary_scores + emerging_scores
+            best_score = max(all_scores) if all_scores else 0.0
+            total_score += best_score
+
+        # Average across all requirements
+        avg_score = total_score / len(requirements)
+
+        # Apply performance modifier
+        performance_modifier = (
+            self.performance.success_rate * 0.3
+            + self.performance.insight_quality_score * 0.7
+        )
+
+        # Weighted final score
+        final_score = avg_score * 0.7 + performance_modifier * 0.3
+
+        return min(max(final_score, 0.0), 1.0)
+
+    def is_compatible_with(self, other_agent_id: str) -> bool:
+        """Check if compatible with another agent."""
+        if other_agent_id in self.incompatible_agents:
+            return False
+
+        # Could add more sophisticated compatibility logic here
+        return True
+
+    def update_performance(self, task_result: Dict[str, Any]) -> None:
+        """Update performance metrics based on task result."""
+        self.performance.total_tasks += 1
+
+        if task_result.get("success", False):
+            self.performance.successful_tasks += 1
+        else:
+            self.performance.failed_tasks += 1
+
+        # Update response time
+        if "response_time_ms" in task_result:
+            # Simple moving average
+            alpha = 0.1  # Learning rate
+            self.performance.average_response_time_ms = (
+                alpha * task_result["response_time_ms"]
+                + (1 - alpha) * self.performance.average_response_time_ms
+            )
+
+        # Update insight metrics
+        if "insights" in task_result:
+            insights = task_result["insights"]
+            self.performance.insights_generated += len(insights)
+
+            # Track unique insights (simple heuristic)
+            unique_count = len(
+                set(insight.get("key", str(i)) for i, insight in enumerate(insights))
+            )
+            self.performance.unique_insights += unique_count
+
+            # Track actionable insights
+            actionable_count = sum(
+                1 for insight in insights if insight.get("actionable", False)
+            )
+            self.performance.actionable_insights += actionable_count
+
+        # Update quality score
+        if "quality_score" in task_result:
+            alpha = 0.1
+            self.performance.average_insight_quality = (
+                alpha * task_result["quality_score"]
+                + (1 - alpha) * self.performance.average_insight_quality
+            )
+
+        self.performance.last_active = datetime.now()
+        self.updated_at = datetime.now()
+
+
+@dataclass
+class Insight:
+    """Individual insight from task execution."""
+
+    insight_id: str = field(default_factory=lambda: str(uuid.uuid4()))
+    content: str = ""
+    insight_type: InsightType = InsightType.ANALYSIS
+    confidence: float = 0.0
+
+    # Quality metrics
+    novelty_score: float = 0.0  # How new/unique is this insight
+    actionability_score: float = 0.0  # How actionable is it
+    impact_score: float = 0.0  # Potential impact if acted upon
+
+    # Metadata
+    generated_by: str = ""  # Agent ID
+    generated_at: datetime = field(default_factory=datetime.now)
+
+    # Related insights
+    builds_on: List[str] = field(default_factory=list)  # IDs of insights this builds on
+    contradicts: List[str] = field(
+        default_factory=list
+    )  # IDs of insights this contradicts
+
+    # Supporting data
+    evidence: List[Dict[str, Any]] = field(default_factory=list)
+    keywords: List[str] = field(default_factory=list)
+
+    @property
+    def quality_score(self) -> float:
+        """Calculate overall quality score."""
+        return (
+            self.confidence * 0.3
+            + self.novelty_score * 0.3
+            + self.actionability_score * 0.3
+            + self.impact_score * 0.1
+        )
+
+    def to_dict(self) -> Dict[str, Any]:
+        """Convert to dictionary."""
+        return {
+            "insight_id": self.insight_id,
+            "content": self.content,
+            "type": self.insight_type.value,
+            "confidence": self.confidence,
+            "quality_score": self.quality_score,
+            "novelty_score": self.novelty_score,
+            "actionability_score": self.actionability_score,
+            "impact_score": self.impact_score,
+            "generated_by": self.generated_by,
+            "generated_at": self.generated_at.isoformat(),
+            "keywords": self.keywords,
+        }
+
+
+@dataclass
+class TaskIteration:
+    """Record of a single task iteration."""
+
+    iteration_number: int
+    started_at: datetime
+    completed_at: Optional[datetime] = None
+
+    # What changed
+    adjustments_made: List[str] = field(default_factory=list)
+    reason_for_iteration: str = ""
+
+    # Results
+    insights_generated: List[Insight] = field(default_factory=list)
+    quality_improvement: float = 0.0  # Change in quality from previous iteration
+
+    # Agent involvement
+    agents_involved: List[str] = field(default_factory=list)
+    consensus_score: float = 0.0
+
+
+@dataclass
+class A2ATask:
+    """
+    Structured task with full lifecycle management.
+
+    Replaces dictionary-based tasks with rich objects that track
+    state transitions, insight collection, and quality metrics.
+    """
+
+    # Identity
+    task_id: str = field(default_factory=lambda: str(uuid.uuid4()))
+    name: str = ""
+    description: str = ""
+
+    # State management
+    state: TaskState = TaskState.CREATED
+    priority: TaskPriority = TaskPriority.MEDIUM
+
+    # Assignment
+    assigned_to: List[str] = field(default_factory=list)  # Agent IDs
+    delegated_by: Optional[str] = None  # Coordinator ID
+
+    # Requirements
+    requirements: List[str] = field(default_factory=list)
+    constraints: Dict[str, Any] = field(default_factory=dict)
+    success_criteria: List[str] = field(default_factory=list)
+
+    # Timeline
+    created_at: datetime = field(default_factory=datetime.now)
+    assigned_at: Optional[datetime] = None
+    started_at: Optional[datetime] = None
+    completed_at: Optional[datetime] = None
+    deadline: Optional[datetime] = None
+
+    # Insights - Primary deliverable
+    insights: List[Insight] = field(default_factory=list)
+
+    # Iterations
+    iterations: List[TaskIteration] = field(default_factory=list)
+    max_iterations: int = 3
+    current_iteration: int = 0
+
+    # Quality tracking
+    target_quality_score: float = 0.85
+    current_quality_score: float = 0.0
+
+    # Context and memory
+    context: Dict[str, Any] = field(default_factory=dict)
+    memory_keys: List[str] = field(default_factory=list)  # Shared memory references
+
+    # Results
+    final_result: Optional[Dict[str, Any]] = None
+    error_message: Optional[str] = None
+
+    # Metadata
+    tags: List[str] = field(default_factory=list)
+    parent_task_id: Optional[str] = None
+    subtask_ids: List[str] = field(default_factory=list)
+
+    def transition_to(self, new_state: TaskState) -> bool:
+        """
+        Transition task to new state with validation.
+
+        Returns True if transition is valid, False otherwise.
+        """
+        valid_transitions = {
+            TaskState.CREATED: [TaskState.ASSIGNED, TaskState.CANCELLED],
+            TaskState.ASSIGNED: [TaskState.IN_PROGRESS, TaskState.CANCELLED],
+            TaskState.IN_PROGRESS: [
+                TaskState.AWAITING_REVIEW,
+                TaskState.FAILED,
+                TaskState.CANCELLED,
+            ],
+            TaskState.AWAITING_REVIEW: [
+                TaskState.ITERATING,
+                TaskState.COMPLETED,
+                TaskState.FAILED,
+            ],
+            TaskState.ITERATING: [
+                TaskState.IN_PROGRESS,
+                TaskState.FAILED,
+                TaskState.CANCELLED,
+            ],
+            TaskState.COMPLETED: [],  # Terminal state
+            TaskState.FAILED: [TaskState.IN_PROGRESS],  # Can retry
+            TaskState.CANCELLED: [],  # Terminal state
+        }
+
+        if new_state not in valid_transitions.get(self.state, []):
+            return False
+
+        # Update timestamps
+        if new_state == TaskState.ASSIGNED:
+            self.assigned_at = datetime.now()
+        elif new_state == TaskState.IN_PROGRESS:
+            if not self.started_at:
+                self.started_at = datetime.now()
+        elif new_state in [TaskState.COMPLETED, TaskState.FAILED, TaskState.CANCELLED]:
+            self.completed_at = datetime.now()
+
+        self.state = new_state
+        return True
+
+    def add_insight(self, insight: Insight) -> None:
+        """Add an insight to the task."""
+        self.insights.append(insight)
+        self._update_quality_score()
+
+    def start_iteration(self, reason: str, adjustments: List[str]) -> TaskIteration:
+        """Start a new iteration of the task."""
+        self.current_iteration += 1
+
+        iteration = TaskIteration(
+            iteration_number=self.current_iteration,
+            started_at=datetime.now(),
+            reason_for_iteration=reason,
+            adjustments_made=adjustments,
+        )
+
+        self.iterations.append(iteration)
+        self.transition_to(TaskState.ITERATING)
+
+        return iteration
+
+    def complete_iteration(
+        self,
+        insights: List[Insight],
+        agents_involved: List[str],
+        consensus_score: float = 0.0,
+    ) -> None:
+        """Complete the current iteration."""
+        if not self.iterations:
+            return
+
+        current = self.iterations[-1]
+        current.completed_at = datetime.now()
+        current.insights_generated = insights
+        current.agents_involved = agents_involved
+        current.consensus_score = consensus_score
+
+        # Calculate quality improvement
+        prev_quality = self.current_quality_score
+        self.insights.extend(insights)
+        self._update_quality_score()
+        current.quality_improvement = self.current_quality_score - prev_quality
+
+        # Transition back to in_progress
+        self.transition_to(TaskState.IN_PROGRESS)
+
+    def _update_quality_score(self) -> None:
+        """Update overall task quality score based on insights."""
+        if not self.insights:
+            self.current_quality_score = 0.0
+            return
+
+        # Average quality of all insights
+        avg_quality = sum(i.quality_score for i in self.insights) / len(self.insights)
+
+        # Bonus for unique insights
+        unique_content = len(set(i.content for i in self.insights))
+        uniqueness_bonus = min(unique_content / len(self.insights), 1.0) * 0.1
+
+        # Bonus for actionable insights
+        actionable_count = sum(1 for i in self.insights if i.actionability_score > 0.7)
+        actionability_bonus = (actionable_count / len(self.insights)) * 0.1
+
+        self.current_quality_score = min(
+            avg_quality + uniqueness_bonus + actionability_bonus, 1.0
+        )
+
+    @property
+    def is_complete(self) -> bool:
+        """Check if task is in a terminal state."""
+        return self.state in [
+            TaskState.COMPLETED,
+            TaskState.FAILED,
+            TaskState.CANCELLED,
+        ]
+
+    @property
+    def needs_iteration(self) -> bool:
+        """Check if task needs another iteration."""
+        return (
+            self.current_quality_score < self.target_quality_score
+            and self.current_iteration < self.max_iterations
+            and self.state == TaskState.AWAITING_REVIEW
+        )
+
+    @property
+    def duration(self) -> Optional[float]:
+        """Get task duration in seconds."""
+        if not self.started_at:
+            return None
+
+        end_time = self.completed_at or datetime.now()
+        return (end_time - self.started_at).total_seconds()
+
+    def to_dict(self) -> Dict[str, Any]:
+        """Convert to dictionary for serialization."""
+        return {
+            "task_id": self.task_id,
+            "name": self.name,
+            "description": self.description,
+            "state": self.state.value,
+            "priority": self.priority.value,
+            "assigned_to": self.assigned_to,
+            "requirements": self.requirements,
+            "created_at": self.created_at.isoformat(),
+            "started_at": self.started_at.isoformat() if self.started_at else None,
+            "completed_at": (
+                self.completed_at.isoformat() if self.completed_at else None
+            ),
+            "insights": [i.to_dict() for i in self.insights],
+            "iterations": len(self.iterations),
+            "current_quality_score": self.current_quality_score,
+            "target_quality_score": self.target_quality_score,
+            "duration": self.duration,
+        }
+
+    @classmethod
+    def from_dict(cls, data: Dict[str, Any]) -> "A2ATask":
+        """Create task from dictionary (backward compatibility)."""
+        # Support old dictionary format
+        if "task_id" not in data:
+            # Legacy format
+            return cls(
+                name=data.get("title", "Unnamed Task"),
+                description=data.get("description", ""),
+                requirements=data.get("requirements", []),
+                context=data,
+            )
+
+        # New format
+        task = cls(
+            task_id=data["task_id"],
+            name=data["name"],
+            description=data.get("description", ""),
+            state=TaskState(data.get("state", "created")),
+            priority=TaskPriority(data.get("priority", "medium")),
+            requirements=data.get("requirements", []),
+        )
+
+        # Restore insights if present
+        if "insights" in data:
+            for insight_data in data["insights"]:
+                insight = Insight(
+                    insight_id=insight_data.get("insight_id", str(uuid.uuid4())),
+                    content=insight_data.get("content", ""),
+                    confidence=insight_data.get("confidence", 0.0),
+                )
+                task.insights.append(insight)
+
+        return task
+
+
+class TaskValidator:
+    """Validates task readiness and quality."""
+
+    @staticmethod
+    def validate_for_assignment(task: A2ATask) -> Tuple[bool, List[str]]:
+        """
+        Validate task is ready for assignment.
+
+        Returns (is_valid, list_of_issues).
+        """
+        issues = []
+
+        if not task.name:
+            issues.append("Task must have a name")
+
+        if not task.description:
+            issues.append("Task must have a description")
+
+        if not task.requirements:
+            issues.append("Task must have at least one requirement")
+
+        if task.state != TaskState.CREATED:
+            issues.append(f"Task must be in CREATED state, not {task.state.value}")
+
+        return len(issues) == 0, issues
+
+    @staticmethod
+    def validate_for_completion(task: A2ATask) -> Tuple[bool, List[str]]:
+        """
+        Validate task is ready for completion.
+
+        Returns (is_valid, list_of_issues).
+        """
+        issues = []
+
+        if not task.insights:
+            issues.append("Task must have at least one insight")
+
+        if task.current_quality_score < task.target_quality_score:
+            issues.append(
+                f"Quality score {task.current_quality_score:.2f} "
+                f"below target {task.target_quality_score:.2f}"
+            )
+
+        if task.state != TaskState.AWAITING_REVIEW:
+            issues.append(
+                f"Task must be in AWAITING_REVIEW state, not {task.state.value}"
+            )
+
+        # Check success criteria
+        # This would need more sophisticated checking in practice
+        if task.success_criteria:
+            issues.append("Success criteria validation not yet implemented")
+
+        return len(issues) == 0, issues
+
+
+# Factory functions for common agent types
+
+
+def create_research_agent_card(agent_id: str, agent_name: str) -> A2AAgentCard:
+    """Create a card for a research-focused agent."""
+    return A2AAgentCard(
+        agent_id=agent_id,
+        agent_name=agent_name,
+        agent_type="research",
+        version="1.0.0",
+        primary_capabilities=[
+            Capability(
+                name="information_retrieval",
+                domain="research",
+                level=CapabilityLevel.EXPERT,
+                description="Expert at finding and synthesizing information from multiple sources",
+                keywords=["search", "retrieval", "synthesis", "analysis"],
+                examples=[
+                    "literature review",
+                    "market research",
+                    "competitive analysis",
+                ],
+            ),
+            Capability(
+                name="data_analysis",
+                domain="research",
+                level=CapabilityLevel.ADVANCED,
+                description="Analyzes complex datasets to extract insights",
+                keywords=["statistics", "patterns", "trends", "visualization"],
+            ),
+        ],
+        secondary_capabilities=[
+            Capability(
+                name="report_generation",
+                domain="documentation",
+                level=CapabilityLevel.ADVANCED,
+                description="Creates comprehensive research reports",
+                keywords=["writing", "documentation", "summaries"],
+            ),
+        ],
+        collaboration_style=CollaborationStyle.COOPERATIVE,
+        description="Specialized in comprehensive research and information synthesis",
+        tags=["research", "analysis", "documentation"],
+    )
+
+
+def create_coding_agent_card(agent_id: str, agent_name: str) -> A2AAgentCard:
+    """Create a card for a coding-focused agent."""
+    return A2AAgentCard(
+        agent_id=agent_id,
+        agent_name=agent_name,
+        agent_type="coding",
+        version="1.0.0",
+        primary_capabilities=[
+            Capability(
+                name="code_generation",
+                domain="software_development",
+                level=CapabilityLevel.EXPERT,
+                description="Generates high-quality code in multiple languages",
+                keywords=["python", "javascript", "java", "implementation"],
+                examples=["API implementation", "algorithm design", "refactoring"],
+            ),
+            Capability(
+                name="debugging",
+                domain="software_development",
+                level=CapabilityLevel.ADVANCED,
+                description="Identifies and fixes bugs in complex codebases",
+                keywords=["troubleshooting", "error", "fix", "debug"],
+            ),
+        ],
+        secondary_capabilities=[
+            Capability(
+                name="code_review",
+                domain="software_development",
+                level=CapabilityLevel.ADVANCED,
+                description="Reviews code for quality, security, and best practices",
+                keywords=["review", "quality", "standards", "security"],
+            ),
+        ],
+        collaboration_style=CollaborationStyle.INDEPENDENT,
+        description="Expert software developer focused on code quality and implementation",
+        tags=["coding", "development", "debugging"],
+    )
+
+
+def create_qa_agent_card(agent_id: str, agent_name: str) -> A2AAgentCard:
+    """Create a card for a QA/testing-focused agent."""
+    return A2AAgentCard(
+        agent_id=agent_id,
+        agent_name=agent_name,
+        agent_type="qa_testing",
+        version="1.0.0",
+        primary_capabilities=[
+            Capability(
+                name="test_design",
+                domain="quality_assurance",
+                level=CapabilityLevel.EXPERT,
+                description="Designs comprehensive test scenarios and edge cases",
+                keywords=["testing", "scenarios", "edge cases", "coverage"],
+                examples=["integration tests", "security tests", "performance tests"],
+            ),
+            Capability(
+                name="bug_detection",
+                domain="quality_assurance",
+                level=CapabilityLevel.EXPERT,
+                description="Identifies defects and quality issues systematically",
+                keywords=["bugs", "defects", "issues", "validation"],
+            ),
+        ],
+        collaboration_style=CollaborationStyle.SUPPORT,
+        description="Quality assurance specialist focused on comprehensive testing",
+        tags=["qa", "testing", "quality", "validation"],
+    )
+
+
+def create_research_task(
+    name: str,
+    description: str,
+    requirements: List[str],
+    priority: TaskPriority = TaskPriority.MEDIUM,
+) -> A2ATask:
+    """Create a research-oriented task."""
+    return A2ATask(
+        name=name,
+        description=description,
+        requirements=requirements,
+        priority=priority,
+        tags=["research", "analysis"],
+        target_quality_score=0.85,
+        max_iterations=3,
+    )
+
+
+def create_implementation_task(
+    name: str,
+    description: str,
+    requirements: List[str],
+    priority: TaskPriority = TaskPriority.HIGH,
+) -> A2ATask:
+    """Create an implementation-oriented task."""
+    return A2ATask(
+        name=name,
+        description=description,
+        requirements=requirements,
+        priority=priority,
+        tags=["implementation", "coding"],
+        target_quality_score=0.90,
+        max_iterations=2,
+    )
+
+
+def create_validation_task(
+    name: str,
+    description: str,
+    requirements: List[str],
+    parent_task_id: str,
+) -> A2ATask:
+    """Create a validation/testing task."""
+    return A2ATask(
+        name=name,
+        description=description,
+        requirements=requirements,
+        priority=TaskPriority.HIGH,
+        parent_task_id=parent_task_id,
+        tags=["validation", "testing"],
+        target_quality_score=0.95,
+        max_iterations=1,
+    )
+
+
+# ============================================================================
+# END OF ENHANCED A2A COMPONENTS
+# ============================================================================
+
 
 @register_node()
 class SharedMemoryPoolNode(Node):
@@ -1043,9 +2017,68 @@ Focus on actionable intelligence rather than just listing what each agent said."
         agent_id: str,
         original_kwargs: Dict[str, Any],
     ) -> List[Dict[str, Any]]:
-        """Use LLM to extract and analyze insights from the response."""
+        """
+        Multi-stage LLM pipeline for high-quality insight extraction.
+
+        This enhanced method implements a 6-stage pipeline as per the A2A enhancement plan:
+        1. Primary extraction with structured output
+        2. Novelty detection against memory pool
+        3. Quality enhancement and validation
+        4. Cross-model validation for reliability (if enabled)
+        5. Impact scoring and ranking
+        6. Meta-insight synthesis
+        """
 
-        # Prepare a focused prompt for insight extraction
+        # Stage 1: Primary LLM extraction with structured output
+        primary_insights = self._stage1_primary_extraction(
+            response, agent_role, original_kwargs
+        )
+
+        if not primary_insights:
+            # Fallback to rule-based extraction
+            return self._extract_insights(response, agent_role)
+
+        # Stage 2: Novelty detection against memory pool
+        memory_pool = original_kwargs.get("memory_pool")
+        if memory_pool:
+            primary_insights = self._stage2_novelty_detection(
+                primary_insights, agent_id, memory_pool
+            )
+
+        # Stage 3: Quality enhancement and validation
+        enhanced_insights = self._stage3_quality_enhancement(
+            primary_insights, agent_role, original_kwargs
+        )
+
+        # Stage 4: Cross-model validation (optional, based on settings)
+        if original_kwargs.get("enable_cross_validation", False):
+            enhanced_insights = self._stage4_cross_model_validation(
+                enhanced_insights, original_kwargs
+            )
+
+        # Stage 5: Impact scoring and ranking
+        scored_insights = self._stage5_impact_scoring(
+            enhanced_insights, agent_role, original_kwargs
+        )
+
+        # Stage 6: Meta-insight synthesis (if multiple high-quality insights)
+        if len(scored_insights) >= 3:
+            meta_insights = self._stage6_meta_insight_synthesis(
+                scored_insights, agent_role, original_kwargs
+            )
+            scored_insights.extend(meta_insights)
+
+        # Sort by quality and return top insights
+        scored_insights.sort(
+            key=lambda x: x.get("quality_score", x.get("importance", 0.5)), reverse=True
+        )
+
+        return scored_insights[:5]  # Return top 5 insights
+
+    def _stage1_primary_extraction(
+        self, response: str, agent_role: str, kwargs: Dict[str, Any]
+    ) -> List[Dict[str, Any]]:
+        """Stage 1: Primary LLM extraction with structured output."""
         insight_extraction_prompt = f"""You are an AI insight extraction specialist. Analyze the following response and extract the most important insights.
 
 Agent Role: {agent_role}
@@ -1054,33 +2087,32 @@ Original Response:
 
 Extract 3-5 key insights from this response. For each insight:
 1. Summarize the core finding or conclusion (max 100 words)
-2. Assign an importance score (0.0-1.0) based on:
-   - Novelty and uniqueness (0.3 weight)
-   - Impact on decision-making (0.4 weight)
-   - Supporting evidence quality (0.3 weight)
-3. Categorize the insight type: finding, conclusion, comparison, recommendation, problem, metric, or pattern
+2. Assign a confidence score (0.0-1.0) based on evidence strength
+3. Categorize the insight type: discovery, analysis, recommendation, warning, opportunity, pattern, or anomaly
 4. Extract key entities mentioned (products, technologies, metrics, etc.)
-5. Suggest relevant tags for categorization
+5. Identify the actionability level (0.0-1.0) - how easy is it to act on this insight?
+6. Note any prerequisites or dependencies
 
 Output your analysis as a JSON array with this structure:
 [
   {{
     "content": "The core insight summarized concisely",
-    "importance": 0.85,
-    "type": "finding",
-    "entities": ["MacBook Air M3", "M2", "battery life"],
-    "tags": ["performance", "comparison", "hardware"],
-    "evidence": "Brief supporting evidence from the text"
+    "confidence": 0.85,
+    "type": "discovery",
+    "entities": ["Entity1", "Entity2"],
+    "actionability": 0.7,
+    "prerequisites": ["Need access to X", "Requires Y"],
+    "evidence": "Brief supporting evidence from the text",
+    "keywords": ["keyword1", "keyword2"]
   }}
 ]
 
 Focus on insights that would be valuable for other agents to know. Ensure the JSON is valid."""
 
         try:
-            # Create a sub-call to the LLM for insight extraction
             extraction_kwargs = {
-                "provider": original_kwargs.get("provider", "ollama"),
-                "model": original_kwargs.get("model", "mistral"),
+                "provider": kwargs.get("provider", "ollama"),
+                "model": kwargs.get("model", "mistral"),
                 "temperature": 0.3,  # Lower temperature for more focused extraction
                 "messages": [
                     {
@@ -1089,10 +2121,9 @@ Focus on insights that would be valuable for other agents to know. Ensure the JS
                     },
                     {"role": "user", "content": insight_extraction_prompt},
                 ],
-                "max_tokens": original_kwargs.get("max_tokens", 1000),
+                "max_tokens": kwargs.get("max_tokens", 1000),
             }
 
-            # Execute LLM call for insight extraction
             extraction_result = super().run(**extraction_kwargs)
 
             if extraction_result.get("success"):
@@ -1100,45 +2131,367 @@ Focus on insights that would be valuable for other agents to know. Ensure the JS
                     "content", ""
                 )
 
-                # Parse the JSON response
+                # Parse JSON response
                 import json
                 import re
 
-                # Try to extract JSON from the response
                 json_match = re.search(r"\[.*?\]", extracted_content, re.DOTALL)
                 if json_match:
                     try:
                         extracted_insights = json.loads(json_match.group())
 
-                        # Convert to our insight format
+                        # Convert to enhanced format
                         insights = []
-                        for item in extracted_insights[:5]:  # Limit to 5 insights
+                        for item in extracted_insights[:5]:
                             insight = {
                                 "content": item.get("content", ""),
-                                "importance": min(
-                                    max(item.get("importance", 0.5), 0.0), 1.0
+                                "confidence": item.get("confidence", 0.5),
+                                "insight_type": InsightType(
+                                    item.get("type", "analysis").upper()
+                                    if item.get("type", "").upper()
+                                    in [e.value.upper() for e in InsightType]
+                                    else "ANALYSIS"
                                 ),
-                                "tags": item.get("tags", []) + [agent_role],
-                                "segment": f"{agent_role}_{item.get('type', 'general')}",
-                                "metadata": {
-                                    "insight_type": item.get("type", "general"),
-                                    "extracted_entities": item.get("entities", []),
-                                    "evidence": item.get("evidence", ""),
-                                    "llm_extracted": True,
-                                },
+                                "entities": item.get("entities", []),
+                                "actionability_score": item.get("actionability", 0.5),
+                                "prerequisites": item.get("prerequisites", []),
+                                "evidence": item.get("evidence", ""),
+                                "keywords": item.get("keywords", []),
+                                "stage": "primary_extraction",
                             }
                             insights.append(insight)
 
                         return insights
-                    except json.JSONDecodeError:
+                    except (json.JSONDecodeError, ValueError):
+                        pass
+        except Exception:
+            pass
+
+        return []
+
+    def _stage2_novelty_detection(
+        self, insights: List[Dict[str, Any]], agent_id: str, memory_pool: Any
+    ) -> List[Dict[str, Any]]:
+        """Stage 2: Novelty detection against memory pool."""
+        for insight in insights:
+            # Search for similar insights in memory
+            similar_memories = memory_pool.execute(
+                action="read",
+                agent_id=agent_id,
+                attention_filter={
+                    "tags": insight.get("keywords", []),
+                    "window_size": 50,  # Check last 50 memories
+                },
+            ).get("memories", [])
+
+            # Calculate novelty score
+            novelty_score = 1.0
+            for memory in similar_memories:
+                # Simple similarity check (could be enhanced with embeddings)
+                memory_content = memory.get("content", "").lower()
+                insight_content = insight["content"].lower()
+
+                # Check for significant overlap
+                common_words = set(memory_content.split()) & set(
+                    insight_content.split()
+                )
+                if len(common_words) > len(insight_content.split()) * 0.5:
+                    novelty_score *= 0.7  # Reduce novelty if similar exists
+
+            insight["novelty_score"] = max(novelty_score, 0.1)
+            insight["similar_insights_count"] = len(similar_memories)
+
+        return insights
+
+    def _stage3_quality_enhancement(
+        self, insights: List[Dict[str, Any]], agent_role: str, kwargs: Dict[str, Any]
+    ) -> List[Dict[str, Any]]:
+        """Stage 3: Quality enhancement and validation."""
+        if not insights:
+            return insights
+
+        # Create a consolidated prompt for quality enhancement
+        insights_text = "\n".join(
+            [
+                f"{i+1}. {insight['content']} (Confidence: {insight['confidence']:.2f})"
+                for i, insight in enumerate(insights)
+            ]
+        )
+
+        enhancement_prompt = f"""As a quality assurance specialist, enhance these insights:
+
+Agent Role: {agent_role}
+Raw Insights:
+{insights_text}
+
+For each insight:
+1. Clarify any ambiguous statements
+2. Add specific metrics or quantities where possible
+3. Identify potential impacts (business, technical, strategic)
+4. Suggest follow-up actions
+5. Rate the overall quality (0.0-1.0)
+
+Respond with a JSON array matching the input order:
+[
+  {{
+    "enhanced_content": "Clearer, more specific version of the insight",
+    "impact": "Description of potential impact",
+    "follow_up_actions": ["Action 1", "Action 2"],
+    "quality_score": 0.85
+  }}
+]"""
+
+        try:
+            enhancement_kwargs = {
+                "provider": kwargs.get("provider", "ollama"),
+                "model": kwargs.get("model", "mistral"),
+                "temperature": 0.2,  # Even lower for enhancement
+                "messages": [
+                    {
+                        "role": "system",
+                        "content": "You are an expert at enhancing and clarifying insights. Always respond with valid JSON.",
+                    },
+                    {"role": "user", "content": enhancement_prompt},
+                ],
+                "max_tokens": 800,
+            }
+
+            enhancement_result = super().run(**enhancement_kwargs)
+
+            if enhancement_result.get("success"):
+                enhanced_content = enhancement_result.get("response", {}).get(
+                    "content", ""
+                )
+
+                import json
+                import re
+
+                json_match = re.search(r"\[.*?\]", enhanced_content, re.DOTALL)
+                if json_match:
+                    try:
+                        enhancements = json.loads(json_match.group())
+
+                        # Merge enhancements with original insights
+                        for i, enhancement in enumerate(enhancements[: len(insights)]):
+                            if enhancement.get("enhanced_content"):
+                                insights[i]["content"] = enhancement["enhanced_content"]
+                            insights[i]["impact_score"] = enhancement.get(
+                                "quality_score", 0.5
+                            )
+                            insights[i]["impact_description"] = enhancement.get(
+                                "impact", ""
+                            )
+                            insights[i]["follow_up_actions"] = enhancement.get(
+                                "follow_up_actions", []
+                            )
+                            insights[i]["stage"] = "quality_enhanced"
+                    except (json.JSONDecodeError, ValueError):
                         pass
+        except Exception:
+            pass
+
+        return insights
+
+    def _stage4_cross_model_validation(
+        self, insights: List[Dict[str, Any]], kwargs: Dict[str, Any]
+    ) -> List[Dict[str, Any]]:
+        """Stage 4: Cross-model validation for reliability (optional)."""
+        # This would validate insights using a different model
+        # For now, we'll simulate by adjusting confidence based on consistency
 
+        alternate_model = kwargs.get("validation_model", kwargs.get("model"))
+        if alternate_model == kwargs.get("model"):
+            # Same model, skip validation
+            return insights
+
+        # In a real implementation, we would re-validate with alternate model
+        # For now, apply a validation factor
+        for insight in insights:
+            insight["cross_validated"] = True
+            insight["confidence"] *= 0.95  # Slight confidence adjustment
+
+        return insights
+
+    def _stage5_impact_scoring(
+        self, insights: List[Dict[str, Any]], agent_role: str, kwargs: Dict[str, Any]
+    ) -> List[Dict[str, Any]]:
+        """Stage 5: Impact scoring and ranking."""
+        for insight in insights:
+            # Calculate comprehensive quality score
+            confidence = insight.get("confidence", 0.5)
+            novelty = insight.get("novelty_score", 0.5)
+            actionability = insight.get("actionability_score", 0.5)
+            impact = insight.get("impact_score", 0.5)
+
+            # Weighted quality score
+            quality_score = (
+                confidence * 0.3 + novelty * 0.3 + actionability * 0.3 + impact * 0.1
+            )
+
+            # Convert to final format
+            insight_obj = Insight(
+                content=insight["content"],
+                insight_type=insight.get("insight_type", InsightType.ANALYSIS),
+                confidence=confidence,
+                novelty_score=novelty,
+                actionability_score=actionability,
+                impact_score=impact,
+                generated_by=kwargs.get("agent_id", ""),
+                keywords=insight.get("keywords", []),
+                evidence=(
+                    [{"text": insight.get("evidence", "")}]
+                    if insight.get("evidence")
+                    else []
+                ),
+            )
+
+            # Add to format expected by memory pool
+            formatted_insight = {
+                "content": insight_obj.content,
+                "importance": quality_score,
+                "quality_score": quality_score,
+                "tags": insight.get("keywords", []) + [agent_role],
+                "segment": f"{agent_role}_{insight_obj.insight_type.value}",
+                "metadata": {
+                    "insight_type": insight_obj.insight_type.value,
+                    "extracted_entities": insight.get("entities", []),
+                    "evidence": insight.get("evidence", ""),
+                    "llm_extracted": True,
+                    "multi_stage_pipeline": True,
+                    "stages_completed": insight.get("stage", "unknown"),
+                    "novelty_score": novelty,
+                    "actionability_score": actionability,
+                    "impact_score": impact,
+                    "quality_score": quality_score,
+                },
+            }
+
+            # Copy over the insight object for reference
+            formatted_insight["insight_object"] = insight_obj
+
+            insights[insights.index(insight)] = formatted_insight
+
+        return insights
+
+    def _stage6_meta_insight_synthesis(
+        self, insights: List[Dict[str, Any]], agent_role: str, kwargs: Dict[str, Any]
+    ) -> List[Dict[str, Any]]:
+        """Stage 6: Meta-insight synthesis from multiple insights."""
+        if len(insights) < 3:
+            return []
+
+        # Create synthesis prompt
+        insights_summary = "\n".join(
+            [
+                f"- {insight['content']} (Quality: {insight.get('quality_score', 0.5):.2f})"
+                for insight in insights[:5]
+            ]
+        )
+
+        synthesis_prompt = f"""Analyze these insights collectively to identify meta-patterns:
+
+Agent Role: {agent_role}
+Individual Insights:
+{insights_summary}
+
+Identify:
+1. Common themes or patterns across insights
+2. Potential synergies or connections
+3. Contradictions or tensions
+4. Emergent conclusions from the collective insights
+
+Provide 1-2 meta-insights that capture higher-level understanding.
+
+Respond with JSON:
+[
+  {{
+    "meta_insight": "Higher-level insight derived from patterns",
+    "supporting_insights": [1, 2, 3],
+    "insight_type": "pattern",
+    "confidence": 0.8
+  }}
+]"""
+
+        try:
+            synthesis_kwargs = {
+                "provider": kwargs.get("provider", "ollama"),
+                "model": kwargs.get("model", "mistral"),
+                "temperature": 0.4,  # Slightly higher for creative synthesis
+                "messages": [
+                    {
+                        "role": "system",
+                        "content": "You are an expert at identifying patterns and synthesizing meta-insights.",
+                    },
+                    {"role": "user", "content": synthesis_prompt},
+                ],
+                "max_tokens": 500,
+            }
+
+            synthesis_result = super().run(**synthesis_kwargs)
+
+            if synthesis_result.get("success"):
+                synthetic_content = synthesis_result.get("response", {}).get(
+                    "content", ""
+                )
+
+                import json
+                import re
+
+                json_match = re.search(r"\[.*?\]", synthetic_content, re.DOTALL)
+                if json_match:
+                    try:
+                        meta_insights_data = json.loads(json_match.group())
+
+                        meta_insights = []
+                        for meta in meta_insights_data[:2]:  # Max 2 meta-insights
+                            meta_insight = Insight(
+                                content=meta.get("meta_insight", ""),
+                                insight_type=InsightType.PATTERN,
+                                confidence=meta.get("confidence", 0.7),
+                                novelty_score=0.9,  # Meta-insights are typically novel
+                                actionability_score=0.6,  # May be less directly actionable
+                                impact_score=0.8,  # But high impact
+                                generated_by=kwargs.get("agent_id", ""),
+                                keywords=["meta-insight", "synthesis", agent_role],
+                            )
+
+                            # Track which insights it builds on
+                            supporting_indices = meta.get("supporting_insights", [])
+                            if supporting_indices:
+                                meta_insight.builds_on = [
+                                    insights[i - 1]
+                                    .get("insight_object", Insight())
+                                    .insight_id
+                                    for i in supporting_indices
+                                    if 0 < i <= len(insights)
+                                ]
+
+                            formatted_meta = {
+                                "content": meta_insight.content,
+                                "importance": meta_insight.quality_score,
+                                "quality_score": meta_insight.quality_score,
+                                "tags": ["meta-insight", "synthesis"] + [agent_role],
+                                "segment": f"{agent_role}_meta_pattern",
+                                "metadata": {
+                                    "insight_type": "meta_pattern",
+                                    "is_meta_insight": True,
+                                    "supporting_insights": supporting_indices,
+                                    "llm_extracted": True,
+                                    "multi_stage_pipeline": True,
+                                    "stage": "meta_synthesis",
+                                },
+                                "insight_object": meta_insight,
+                            }
+
+                            meta_insights.append(formatted_meta)
+
+                        return meta_insights
+                    except (json.JSONDecodeError, ValueError):
+                        pass
         except Exception:
-            # Log the error but don't fail - fall back to rule-based extraction
             pass
 
-        # If LLM extraction fails, fall back to rule-based
-        return self._extract_insights(response, agent_role)
+        return []
 
 
 @register_node()
@@ -1242,6 +2595,12 @@ class A2ACoordinatorNode(CycleAwareNode):
         self.task_queue = deque()
         self.consensus_sessions = {}
 
+        # Enhanced features
+        self.agent_cards: Dict[str, A2AAgentCard] = {}
+        self.active_tasks: Dict[str, A2ATask] = {}
+        self.completed_tasks: List[A2ATask] = []
+        self.task_history_limit = 100  # Keep last 100 completed tasks
+
     def get_parameters(self) -> Dict[str, NodeParameter]:
         return {
             "action": NodeParameter(
@@ -1249,7 +2608,7 @@ class A2ACoordinatorNode(CycleAwareNode):
                 type=str,
                 required=False,
                 default="coordinate",
-                description="Action: 'register', 'delegate', 'broadcast', 'consensus', 'coordinate'",
+                description="Action: 'register', 'register_with_card', 'delegate', 'broadcast', 'consensus', 'coordinate', 'create_task', 'update_task_state', 'get_task_insights', 'match_agents_to_task'",
             ),
             "agent_info": NodeParameter(
                 name="agent_info",
@@ -1257,12 +2616,30 @@ class A2ACoordinatorNode(CycleAwareNode):
                 required=False,
                 description="Information about agent (for registration)",
             ),
+            "agent_id": NodeParameter(
+                name="agent_id",
+                type=str,
+                required=False,
+                description="Unique identifier for an agent (for register_with_card, update_task_state)",
+            ),
+            "agent_card": NodeParameter(
+                name="agent_card",
+                type=dict,
+                required=False,
+                description="Rich capability card for agent registration (for register_with_card action)",
+            ),
             "task": NodeParameter(
                 name="task",
                 type=dict,
                 required=False,
                 description="Task to delegate or coordinate",
             ),
+            "task_id": NodeParameter(
+                name="task_id",
+                type=str,
+                required=False,
+                description="ID of an existing task (for update_task_state, get_task_insights, delegate)",
+            ),
             "message": NodeParameter(
                 name="message",
                 type=dict,
@@ -1289,6 +2666,67 @@ class A2ACoordinatorNode(CycleAwareNode):
                 default="best_match",
                 description="Strategy: 'best_match', 'round_robin', 'broadcast', 'auction'",
             ),
+            "task_type": NodeParameter(
+                name="task_type",
+                type=str,
+                required=False,
+                default="research",
+                description="Type of task to create: 'research', 'implementation', 'validation'",
+            ),
+            "name": NodeParameter(
+                name="name",
+                type=str,
+                required=False,
+                default="",
+                description="Name of the task",
+            ),
+            "description": NodeParameter(
+                name="description",
+                type=str,
+                required=False,
+                default="",
+                description="Description of the task",
+            ),
+            "requirements": NodeParameter(
+                name="requirements",
+                type=list,
+                required=False,
+                default=[],
+                description="List of requirements for the task",
+            ),
+            "priority": NodeParameter(
+                name="priority",
+                type=str,
+                required=False,
+                default="medium",
+                description="Task priority: 'low', 'medium', 'high', 'critical'",
+            ),
+            "new_state": NodeParameter(
+                name="new_state",
+                type=str,
+                required=False,
+                description="New state for task transition",
+            ),
+            "insights": NodeParameter(
+                name="insights",
+                type=list,
+                required=False,
+                default=[],
+                description="List of insights to add to task",
+            ),
+            "min_quality": NodeParameter(
+                name="min_quality",
+                type=float,
+                required=False,
+                default=0.0,
+                description="Minimum quality score for insight filtering",
+            ),
+            "insight_type": NodeParameter(
+                name="insight_type",
+                type=str,
+                required=False,
+                description="Type of insights to filter",
+            ),
         }
 
     def run(self, **kwargs) -> Dict[str, Any]:
@@ -1348,13 +2786,30 @@ class A2ACoordinatorNode(CycleAwareNode):
             coordination_history = prev_state.get("coordination_history", [])
             agent_performance_history = prev_state.get("agent_performance", {})
 
-        # Execute the coordination action
-        if action == "register":
+        # Execute the coordination action - enhanced actions first
+        if action == "register_with_card":
+            result = self._register_agent_with_card(kwargs, context)
+        elif action == "create_task":
+            result = self._create_structured_task(kwargs)
+        elif action == "update_task_state":
+            result = self._update_task_state(kwargs)
+        elif action == "get_task_insights":
+            result = self._get_task_insights(kwargs)
+        elif action == "match_agents_to_task":
+            result = self._match_agents_to_task(kwargs)
+        # Original actions with enhancement support
+        elif action == "register":
             result = self._register_agent(kwargs, context)
         elif action == "delegate":
-            result = self._delegate_task(
-                kwargs, context, coordination_history, agent_performance_history
-            )
+            # Check if we should use enhanced delegation
+            if self.agent_cards or kwargs.get("task_id") in self.active_tasks:
+                result = self._enhanced_delegate_task(
+                    kwargs, context, coordination_history, agent_performance_history
+                )
+            else:
+                result = self._delegate_task(
+                    kwargs, context, coordination_history, agent_performance_history
+                )
         elif action == "broadcast":
             result = self._broadcast_message(kwargs, context)
         elif action == "consensus":
@@ -1427,6 +2882,7 @@ class A2ACoordinatorNode(CycleAwareNode):
         if not agent_id:
             return {"success": False, "error": "Agent ID required"}
 
+        # Create base registration
         self.registered_agents[agent_id] = {
             "id": agent_id,
             "skills": agent_info.get("skills", []),
@@ -1437,6 +2893,12 @@ class A2ACoordinatorNode(CycleAwareNode):
             "success_rate": 1.0,
         }
 
+        # Create default agent card if not exists
+        if agent_id not in self.agent_cards:
+            self.agent_cards[agent_id] = self._create_default_agent_card(
+                agent_id, agent_info.get("skills", [])
+            )
+
         return {
             "success": True,
             "agent_id": agent_id,
@@ -1805,3 +3267,412 @@ class A2ACoordinatorNode(CycleAwareNode):
             return bids[0]["agent"]
 
         return None
+
+    # Enhanced methods to be added to A2ACoordinatorNode
+
+    # =========================================================================
+    # ENHANCED METHODS FOR AGENT CARDS AND TASK MANAGEMENT
+    # =========================================================================
+
+    def _register_agent_with_card(
+        self, kwargs: Dict[str, Any], context: Dict[str, Any]
+    ) -> Dict[str, Any]:
+        """Register an agent with a rich capability card."""
+        agent_id = kwargs.get("agent_id")
+        card_data = kwargs.get("agent_card")
+
+        if not agent_id or not card_data:
+            return {"success": False, "error": "agent_id and agent_card required"}
+
+        # Create or update agent card
+        if isinstance(card_data, dict):
+            card = A2AAgentCard.from_dict(card_data)
+        else:
+            card = card_data
+
+        self.agent_cards[agent_id] = card
+
+        # Also register with base system for compatibility
+        capabilities = [cap.name for cap in card.primary_capabilities]
+        self._register_agent(
+            {
+                "agent_info": {
+                    "id": agent_id,
+                    "skills": capabilities,
+                    "role": card.agent_type,
+                }
+            },
+            context,
+        )
+
+        return {
+            "success": True,
+            "agent_id": agent_id,
+            "capabilities_registered": len(capabilities),
+            "card_version": card.version,
+        }
+
+    def _create_structured_task(self, kwargs: Dict[str, Any]) -> Dict[str, Any]:
+        """Create a structured task with lifecycle management."""
+        task_type = kwargs.get("task_type", "research")
+        name = kwargs.get("name", "")
+        description = kwargs.get("description", "")
+        requirements = kwargs.get("requirements", [])
+        priority = kwargs.get("priority", "medium")
+
+        # Create appropriate task type
+        if task_type == "research":
+            task = create_research_task(
+                name=name,
+                description=description,
+                requirements=requirements,
+                priority=TaskPriority(priority),
+            )
+        elif task_type == "implementation":
+            task = create_implementation_task(
+                name=name,
+                description=description,
+                requirements=requirements,
+                priority=TaskPriority(priority),
+            )
+        elif task_type == "validation":
+            parent_id = kwargs.get("parent_task_id")
+            task = create_validation_task(
+                name=name,
+                description=description,
+                requirements=requirements,
+                parent_task_id=parent_id,
+            )
+        else:
+            # Generic task
+            task = A2ATask(
+                name=name,
+                description=description,
+                requirements=requirements,
+                priority=TaskPriority(priority),
+            )
+
+        # Add any additional context
+        if "context" in kwargs:
+            task.context.update(kwargs["context"])
+
+        # Store task
+        self.active_tasks[task.task_id] = task
+
+        return {"success": True, "task_id": task.task_id, "task": task.to_dict()}
+
+    def _enhanced_delegate_task(
+        self,
+        kwargs: Dict[str, Any],
+        context: Dict[str, Any],
+        coordination_history: List[Dict],
+        agent_performance_history: Dict[str, Dict],
+    ) -> Dict[str, Any]:
+        """Enhanced delegation using agent cards for better matching."""
+        task_id = kwargs.get("task_id")
+        task_dict = kwargs.get("task", {})
+
+        # Check if this is a structured task
+        if task_id and task_id in self.active_tasks:
+            task = self.active_tasks[task_id]
+
+            # Validate task is ready
+            is_valid, issues = TaskValidator.validate_for_assignment(task)
+            if not is_valid:
+                return {
+                    "success": False,
+                    "error": f"Task not ready for assignment: {', '.join(issues)}",
+                }
+
+            # Find best agents using cards
+            best_agents = self._find_best_agents_for_task(task)
+
+            if not best_agents:
+                # Fall back to base delegation
+                return self._delegate_task(
+                    kwargs, context, coordination_history, agent_performance_history
+                )
+
+            # Assign to best agents
+            task.assigned_to = [agent_id for agent_id, _ in best_agents[:3]]
+            task.transition_to(TaskState.ASSIGNED)
+            task.assigned_at = datetime.now()
+
+            # Use first agent for delegation
+            return {
+                "success": True,
+                "delegated_to": task.assigned_to[0],
+                "task_id": task.task_id,
+                "match_score": best_agents[0][1],
+                "state": task.state.value,
+            }
+
+        # Not a structured task, but we can still use agent cards for better matching
+        if self.agent_cards and task_dict.get("required_skills"):
+            # Try to match using agent cards
+            required_skills = task_dict.get("required_skills", [])
+
+            # Find agents that match the requirements
+            matching_agents = []
+            for agent_id, agent_info in self.registered_agents.items():
+                if agent_info["status"] == "available" and agent_id in self.agent_cards:
+                    card = self.agent_cards[agent_id]
+                    # Check if any capability matches the required skills
+                    all_capabilities = list(card.primary_capabilities) + list(
+                        card.secondary_capabilities
+                    )
+                    for cap in all_capabilities:
+                        cap_keywords = getattr(cap, "keywords", [])
+                        for req in required_skills:
+                            if req.lower() in cap.name.lower() or any(
+                                req.lower() in kw.lower() for kw in cap_keywords
+                            ):
+                                matching_agents.append(agent_id)
+                                break
+                        else:
+                            continue
+                        break
+
+            if matching_agents:
+                # Override available_agents with matched agents, adding card capabilities as skills
+                enhanced_agents = []
+                for agent_id in matching_agents:
+                    agent_copy = dict(self.registered_agents[agent_id])
+                    # Add card capabilities as skills for matching
+                    if agent_id in self.agent_cards:
+                        card = self.agent_cards[agent_id]
+                        card_skills = [cap.name for cap in card.primary_capabilities]
+                        # Merge original skills with card capabilities
+                        agent_copy["skills"] = list(
+                            set(agent_copy.get("skills", []) + card_skills)
+                        )
+                    enhanced_agents.append(agent_copy)
+                kwargs["available_agents"] = enhanced_agents
+
+        # Use base delegation which will use the available_agents if provided
+        return self._delegate_task(
+            kwargs, context, coordination_history, agent_performance_history
+        )
+
+    def _find_best_agents_for_task(self, task: A2ATask) -> List[Tuple[str, float]]:
+        """Find best agents for a task using agent cards."""
+        matches = []
+
+        for agent_id, card in self.agent_cards.items():
+            # Skip if incompatible
+            if task.delegated_by and not card.is_compatible_with(task.delegated_by):
+                continue
+
+            # Calculate match score
+            score = card.calculate_match_score(task.requirements)
+
+            # Apply collaboration style bonus
+            if len(task.assigned_to) > 0:
+                if card.collaboration_style == CollaborationStyle.COOPERATIVE:
+                    score *= 1.1
+                elif card.collaboration_style == CollaborationStyle.INDEPENDENT:
+                    score *= 0.9
+
+            # Apply performance history bonus
+            if card.performance.total_tasks > 10:
+                score *= 0.8 + 0.2 * card.performance.success_rate
+
+            matches.append((agent_id, score))
+
+        # Sort by score descending
+        matches.sort(key=lambda x: x[1], reverse=True)
+
+        return matches
+
+    def _update_task_state(self, kwargs: Dict[str, Any]) -> Dict[str, Any]:
+        """Update task state and handle transitions."""
+        task_id = kwargs.get("task_id")
+        new_state = kwargs.get("new_state")
+        insights = kwargs.get("insights", [])
+        agent_id = kwargs.get("agent_id")
+
+        if task_id not in self.active_tasks:
+            return {"success": False, "error": f"Task {task_id} not found"}
+
+        task = self.active_tasks[task_id]
+
+        # Handle state transition
+        if new_state:
+            success = task.transition_to(TaskState(new_state))
+            if not success:
+                return {
+                    "success": False,
+                    "error": f"Invalid transition from {task.state.value} to {new_state}",
+                }
+
+        # Add insights if provided
+        for insight_data in insights:
+            if isinstance(insight_data, dict):
+                insight = Insight(
+                    content=insight_data.get("content", ""),
+                    insight_type=InsightType(insight_data.get("type", "analysis")),
+                    confidence=insight_data.get("confidence", 0.0),
+                    novelty_score=insight_data.get("novelty_score", 0.0),
+                    actionability_score=insight_data.get("actionability_score", 0.0),
+                    impact_score=insight_data.get("impact_score", 0.0),
+                    generated_by=agent_id or "",
+                    keywords=insight_data.get("keywords", []),
+                )
+            else:
+                insight = insight_data
+
+            task.add_insight(insight)
+
+        # Update agent performance if we have cards
+        if agent_id and agent_id in self.agent_cards:
+            card = self.agent_cards[agent_id]
+            card.update_performance(
+                {
+                    "success": task.state != TaskState.FAILED,
+                    "insights": insights,
+                    "quality_score": task.current_quality_score,
+                }
+            )
+
+        # Check if task needs iteration
+        if task.state == TaskState.AWAITING_REVIEW and task.needs_iteration:
+            return {
+                "success": True,
+                "task_state": task.state.value,
+                "needs_iteration": True,
+                "current_quality": task.current_quality_score,
+                "target_quality": task.target_quality_score,
+                "iteration": task.current_iteration + 1,
+            }
+
+        # Move completed tasks to history
+        if task.is_complete:
+            self.completed_tasks.append(task)
+            # Limit history size
+            if len(self.completed_tasks) > self.task_history_limit:
+                self.completed_tasks = self.completed_tasks[-self.task_history_limit :]
+            del self.active_tasks[task_id]
+
+        return {
+            "success": True,
+            "task_state": task.state.value,
+            "quality_score": task.current_quality_score,
+            "insights_count": len(task.insights),
+        }
+
+    def _get_task_insights(self, kwargs: Dict[str, Any]) -> Dict[str, Any]:
+        """Get insights from a task."""
+        task_id = kwargs.get("task_id")
+        min_quality = kwargs.get("min_quality", 0.0)
+        insight_type = kwargs.get("insight_type")
+
+        # Check active tasks
+        if task_id in self.active_tasks:
+            task = self.active_tasks[task_id]
+        # Check completed tasks
+        else:
+            task = next((t for t in self.completed_tasks if t.task_id == task_id), None)
+
+        if not task:
+            return {"success": False, "error": f"Task {task_id} not found"}
+
+        # Filter insights
+        insights = task.insights
+
+        if min_quality > 0:
+            insights = [i for i in insights if i.quality_score >= min_quality]
+
+        if insight_type:
+            type_filter = InsightType(insight_type)
+            insights = [i for i in insights if i.insight_type == type_filter]
+
+        return {
+            "success": True,
+            "task_id": task_id,
+            "task_state": task.state.value,
+            "insights": [i.to_dict() for i in insights],
+            "total_insights": len(task.insights),
+            "filtered_insights": len(insights),
+        }
+
+    def _match_agents_to_task(self, kwargs: Dict[str, Any]) -> Dict[str, Any]:
+        """Match agents to task requirements without delegation."""
+        task_id = kwargs.get("task_id")
+        requirements = kwargs.get("requirements", [])
+
+        # Get task if ID provided
+        if task_id and task_id in self.active_tasks:
+            task = self.active_tasks[task_id]
+            requirements = task.requirements
+        elif not requirements:
+            return {
+                "success": False,
+                "error": "Either task_id or requirements must be provided",
+            }
+
+        # Create temporary task for matching if needed
+        if not task_id:
+            task = A2ATask(requirements=requirements)
+
+        # Find matches
+        matches = self._find_best_agents_for_task(task)
+
+        # Format results
+        agent_matches = []
+        for agent_id, score in matches[:10]:  # Top 10 matches
+            card = self.agent_cards[agent_id]
+            agent_matches.append(
+                {
+                    "agent_id": agent_id,
+                    "agent_name": card.agent_name,
+                    "match_score": score,
+                    "primary_capabilities": [
+                        cap.name for cap in card.primary_capabilities
+                    ],
+                    "performance": {
+                        "success_rate": card.performance.success_rate,
+                        "insight_quality": card.performance.insight_quality_score,
+                    },
+                    "collaboration_style": card.collaboration_style.value,
+                }
+            )
+
+        return {
+            "success": True,
+            "requirements": requirements,
+            "matched_agents": agent_matches,
+            "total_agents": len(self.agent_cards),
+        }
+
+    def _create_default_agent_card(
+        self, agent_id: str, capabilities: List[str]
+    ) -> A2AAgentCard:
+        """Create a basic agent card from capability list."""
+        # Guess agent type from capabilities
+        if any("research" in cap.lower() for cap in capabilities):
+            return create_research_agent_card(agent_id, agent_id)
+        elif any(
+            "code" in cap.lower() or "implement" in cap.lower() for cap in capabilities
+        ):
+            return create_coding_agent_card(agent_id, agent_id)
+        elif any("test" in cap.lower() or "qa" in cap.lower() for cap in capabilities):
+            return create_qa_agent_card(agent_id, agent_id)
+        else:
+            # Generic card
+            return A2AAgentCard(
+                agent_id=agent_id,
+                agent_name=agent_id,
+                agent_type="generic",
+                version="1.0.0",
+                primary_capabilities=[
+                    Capability(
+                        name=cap,
+                        domain="general",
+                        level=CapabilityLevel.INTERMEDIATE,
+                        description=f"Capable of {cap}",
+                        keywords=[cap.lower()],
+                    )
+                    for cap in capabilities[:3]  # Limit to 3 primary
+                ],
+                description=f"Agent with capabilities: {', '.join(capabilities)}",
+            )