feat: add runtime dynamic team controls

backend/agents/dynamic_team_types.py

@@ -0,0 +1,372 @@
+# -*- coding: utf-8 -*-
+"""Dynamic Team Types - Core data types for PM-driven analyst team management.
+
+This module provides data structures for:
+- Analyst persona definitions (custom analyst types)
+- Analyst creation configuration (custom SOUL.md, AGENTS.md, etc.)
+- Dynamic team runtime state tracking
+
+These types enable the Portfolio Manager to dynamically create, clone, and manage
+analyst agents with custom configurations beyond the predefined 4 analyst types.
+"""
+from __future__ import annotations
+
+from dataclasses import dataclass, field
+from typing import Optional, Dict, Any, List
+from datetime import datetime
+
+
+@dataclass
+class AnalystPersona:
+    """Analyst role definition - extends or replaces personas.yaml entries.
+
+    Defines the identity, focus areas, and characteristics of an analyst type.
+    Can be used to create entirely new analyst types at runtime.
+
+    Attributes:
+        name: Display name for the analyst (e.g., "期权策略分析师")
+        focus: List of focus areas (e.g., ["期权定价", "波动率交易"])
+        description: Detailed description of the analyst's role and expertise
+        preferred_tools: Optional list of preferred tool types or categories
+        icon: Optional icon identifier for frontend display
+    """
+    name: str
+    focus: List[str]
+    description: str
+    preferred_tools: Optional[List[str]] = None
+    icon: Optional[str] = None
+
+    def to_dict(self) -> Dict[str, Any]:
+        """Convert to dictionary for serialization."""
+        return {
+            "name": self.name,
+            "focus": self.focus,
+            "description": self.description,
+            "preferred_tools": self.preferred_tools,
+            "icon": self.icon,
+        }
+
+    @classmethod
+    def from_dict(cls, data: Dict[str, Any]) -> AnalystPersona:
+        """Create from dictionary."""
+        return cls(
+            name=data["name"],
+            focus=data.get("focus", []),
+            description=data.get("description", ""),
+            preferred_tools=data.get("preferred_tools"),
+            icon=data.get("icon"),
+        )
+
+
+@dataclass
+class AnalystConfig:
+    """Complete configuration for dynamically creating an analyst.
+
+    This dataclass allows the PM to specify all aspects of analyst creation,
+    including custom workspace files, model overrides, and skill selections.
+
+    Attributes:
+        persona: Complete persona definition (if creating custom type)
+        analyst_type: Reference to predefined type (e.g., "technical_analyst")
+        soul_md: Custom SOUL.md content (overrides default generation)
+        agents_md: Custom AGENTS.md content (overrides default generation)
+        profile_md: Custom PROFILE.md content (overrides default generation)
+        skills: List of skill IDs to enable for this analyst
+        model_name: Override default model for this analyst
+        memory_config: Custom memory system configuration
+        tags: Classification tags (e.g., ["options", "derivatives"])
+        parent_id: If cloned, the source analyst ID
+    """
+    # Identity configuration
+    persona: Optional[AnalystPersona] = None
+    analyst_type: Optional[str] = None  # Reference to predefined type
+
+    # Workspace file contents (override default generation)
+    soul_md: Optional[str] = None
+    agents_md: Optional[str] = None
+    profile_md: Optional[str] = None
+    bootstrap_md: Optional[str] = None
+
+    # Runtime configuration
+    skills: Optional[List[str]] = field(default_factory=list)
+    model_name: Optional[str] = None
+    memory_config: Optional[Dict[str, Any]] = field(default_factory=dict)
+
+    # Metadata
+    tags: Optional[List[str]] = field(default_factory=list)
+    parent_id: Optional[str] = None  # For clone tracking
+
+    def __post_init__(self):
+        """Initialize default collections."""
+        if self.skills is None:
+            self.skills = []
+        if self.memory_config is None:
+            self.memory_config = {}
+        if self.tags is None:
+            self.tags = []
+
+    def to_dict(self) -> Dict[str, Any]:
+        """Convert to dictionary for serialization."""
+        return {
+            "persona": self.persona.to_dict() if self.persona else None,
+            "analyst_type": self.analyst_type,
+            "soul_md": self.soul_md,
+            "agents_md": self.agents_md,
+            "profile_md": self.profile_md,
+            "bootstrap_md": self.bootstrap_md,
+            "skills": self.skills,
+            "model_name": self.model_name,
+            "memory_config": self.memory_config,
+            "tags": self.tags,
+            "parent_id": self.parent_id,
+        }
+
+    @classmethod
+    def from_dict(cls, data: Dict[str, Any]) -> AnalystConfig:
+        """Create from dictionary."""
+        persona_data = data.get("persona")
+        return cls(
+            persona=AnalystPersona.from_dict(persona_data) if persona_data else None,
+            analyst_type=data.get("analyst_type"),
+            soul_md=data.get("soul_md"),
+            agents_md=data.get("agents_md"),
+            profile_md=data.get("profile_md"),
+            bootstrap_md=data.get("bootstrap_md"),
+            skills=data.get("skills", []),
+            model_name=data.get("model_name"),
+            memory_config=data.get("memory_config", {}),
+            tags=data.get("tags", []),
+            parent_id=data.get("parent_id"),
+        )
+
+    def get_effective_analyst_type(self) -> Optional[str]:
+        """Get the effective analyst type for tool selection.
+
+        Returns analyst_type if set, otherwise derives from persona name.
+        """
+        if self.analyst_type:
+            return self.analyst_type
+        if self.persona:
+            # Derive type ID from persona name (e.g., "期权策略分析师" -> "options_strategist")
+            return self._derive_type_id(self.persona.name)
+        return None
+
+    @staticmethod
+    def _derive_type_id(name: str) -> str:
+        """Derive a type ID from a display name."""
+        import re
+        # Convert Chinese or mixed names to snake_case
+        # Remove special characters, keep alphanumeric and spaces
+        cleaned = re.sub(r'[^\w\s]', '', name)
+        # Convert to lowercase and replace spaces with underscores
+        return cleaned.lower().strip().replace(' ', '_')
+
+
+@dataclass
+class DynamicAnalystInstance:
+    """Runtime information about a dynamically created analyst.
+
+    Tracks the creation metadata and current state of a dynamic analyst.
+
+    Attributes:
+        agent_id: Unique identifier for this analyst instance
+        config: The configuration used to create this analyst
+        created_at: Timestamp when the analyst was created
+        created_by: Identifier of the agent that created this analyst (usually PM)
+        status: Current status (active, paused, removed)
+    """
+    agent_id: str
+    config: AnalystConfig
+    created_at: str = field(default_factory=lambda: datetime.now().isoformat())
+    created_by: str = "portfolio_manager"
+    status: str = "active"  # active, paused, removed
+
+    def to_dict(self) -> Dict[str, Any]:
+        """Convert to dictionary for serialization."""
+        return {
+            "agent_id": self.agent_id,
+            "config": self.config.to_dict(),
+            "created_at": self.created_at,
+            "created_by": self.created_by,
+            "status": self.status,
+        }
+
+    @classmethod
+    def from_dict(cls, data: Dict[str, Any]) -> DynamicAnalystInstance:
+        """Create from dictionary."""
+        return cls(
+            agent_id=data["agent_id"],
+            config=AnalystConfig.from_dict(data.get("config", {})),
+            created_at=data.get("created_at", datetime.now().isoformat()),
+            created_by=data.get("created_by", "portfolio_manager"),
+            status=data.get("status", "active"),
+        )
+
+
+@dataclass
+class DynamicTeamState:
+    """Complete runtime state for dynamic analyst team management.
+
+    This state is persisted alongside TEAM_PIPELINE.yaml and tracks:
+    - Custom analyst types registered at runtime
+    - All dynamically created analyst instances
+    - Configuration snapshots for cloning
+
+    Attributes:
+        run_id: The run configuration this state belongs to
+        registered_types: Runtime-registered analyst type definitions
+        instances: Dynamically created analyst instances
+        version: State format version for migration handling
+    """
+    run_id: str
+    registered_types: Dict[str, AnalystPersona] = field(default_factory=dict)
+    instances: Dict[str, DynamicAnalystInstance] = field(default_factory=dict)
+    version: int = 1
+
+    def register_type(self, type_id: str, persona: AnalystPersona) -> bool:
+        """Register a new analyst type.
+
+        Returns:
+            True if registered, False if type_id already exists
+        """
+        if type_id in self.registered_types:
+            return False
+        self.registered_types[type_id] = persona
+        return True
+
+    def add_instance(self, instance: DynamicAnalystInstance) -> None:
+        """Add a new analyst instance."""
+        self.instances[instance.agent_id] = instance
+
+    def remove_instance(self, agent_id: str) -> bool:
+        """Mark an instance as removed.
+
+        Returns:
+            True if instance was found and removed
+        """
+        if agent_id in self.instances:
+            self.instances[agent_id].status = "removed"
+            return True
+        return False
+
+    def get_active_instances(self) -> List[DynamicAnalystInstance]:
+        """Get all active (non-removed) analyst instances."""
+        return [
+            inst for inst in self.instances.values()
+            if inst.status == "active"
+        ]
+
+    def to_dict(self) -> Dict[str, Any]:
+        """Convert to dictionary for serialization."""
+        return {
+            "run_id": self.run_id,
+            "registered_types": {
+                k: v.to_dict() for k, v in self.registered_types.items()
+            },
+            "instances": {
+                k: v.to_dict() for k, v in self.instances.items()
+            },
+            "version": self.version,
+        }
+
+    @classmethod
+    def from_dict(cls, data: Dict[str, Any]) -> DynamicTeamState:
+        """Create from dictionary."""
+        registered_types = {
+            k: AnalystPersona.from_dict(v)
+            for k, v in data.get("registered_types", {}).items()
+        }
+        instances = {
+            k: DynamicAnalystInstance.from_dict(v)
+            for k, v in data.get("instances", {}).items()
+        }
+        return cls(
+            run_id=data.get("run_id", "unknown"),
+            registered_types=registered_types,
+            instances=instances,
+            version=data.get("version", 1),
+        )
+
+
+@dataclass
+class CreateAnalystResult:
+    """Result of creating a dynamic analyst.
+
+    Attributes:
+        success: Whether creation was successful
+        agent_id: The ID of the created analyst (if successful)
+        message: Human-readable result message
+        error: Error details (if failed)
+    """
+    success: bool
+    agent_id: Optional[str] = None
+    message: str = ""
+    error: Optional[str] = None
+
+    def to_dict(self) -> Dict[str, Any]:
+        """Convert to dictionary for API responses."""
+        return {
+            "success": self.success,
+            "agent_id": self.agent_id,
+            "message": self.message,
+            "error": self.error,
+        }
+
+
+@dataclass
+class CloneAnalystRequest:
+    """Request to clone an existing analyst.
+
+    Attributes:
+        source_id: ID of the analyst to clone
+        new_id: ID for the new analyst
+        config_overrides: Configuration fields to override
+    """
+    source_id: str
+    new_id: str
+    config_overrides: Optional[Dict[str, Any]] = field(default_factory=dict)
+
+    def __post_init__(self):
+        if self.config_overrides is None:
+            self.config_overrides = {}
+
+
+@dataclass
+class AnalystTypeInfo:
+    """Information about an available analyst type.
+
+    Used for listing all available types (predefined + runtime-registered).
+
+    Attributes:
+        type_id: Unique identifier for this type
+        name: Display name
+        description: Type description
+        is_builtin: Whether this is a built-in type or runtime-registered
+        source: Source of this type (e.g., "constants", "runtime", "config")
+    """
+    type_id: str
+    name: str
+    description: str
+    is_builtin: bool
+    source: str
+
+    def to_dict(self) -> Dict[str, Any]:
+        """Convert to dictionary for API responses."""
+        return {
+            "type_id": self.type_id,
+            "name": self.name,
+            "description": self.description,
+            "is_builtin": self.is_builtin,
+            "source": self.source,
+        }
+
+
+__all__ = [
+    "AnalystPersona",
+    "AnalystConfig",
+    "DynamicAnalystInstance",
+    "DynamicTeamState",
+    "CreateAnalystResult",
+    "CloneAnalystRequest",
+    "AnalystTypeInfo",
+]

backend/agents/toolkit_factory.py

@@ -14,6 +14,14 @@ from backend.agents.agent_workspace import load_agent_workspace_config
 from backend.agents.skills_manager import SkillsManager
 from backend.agents.skill_metadata import parse_skill_metadata
 from backend.config.bootstrap_config import get_bootstrap_config_for_run
+from backend.tools.dynamic_team_tools import (
+    create_analyst,
+    clone_analyst,
+    remove_analyst,
+    list_analyst_types,
+    get_analyst_info,
+    get_team_summary,
+)
 
 
 def load_agent_profiles() -> Dict[str, Dict[str, Any]]:
@@ -138,6 +146,23 @@ def _register_portfolio_tool_groups(toolkit: Any, pm_agent: Any) -> None:
             group_name="portfolio_ops",
         )
 
+    # Register dynamic team management tools
+    toolkit.create_tool_group(
+        group_name="dynamic_team",
+        description="Dynamic analyst team management tools.",
+        active=False,
+        notes=(
+            "Use these tools to create, clone, and manage analyst agents dynamically. "
+            "Only available when allow_dynamic_team_update is enabled."
+        ),
+    )
+    toolkit.register_tool_function(create_analyst, group_name="dynamic_team")
+    toolkit.register_tool_function(clone_analyst, group_name="dynamic_team")
+    toolkit.register_tool_function(remove_analyst, group_name="dynamic_team")
+    toolkit.register_tool_function(list_analyst_types, group_name="dynamic_team")
+    toolkit.register_tool_function(get_analyst_info, group_name="dynamic_team")
+    toolkit.register_tool_function(get_team_summary, group_name="dynamic_team")
+
 
 def _register_risk_tool_groups(toolkit: Any) -> None:
     """注册风险工具组"""

backend/api/__init__.py

@@ -13,6 +13,7 @@ from .workspaces import router as workspaces_router
 from .guard import router as guard_router
 from .runtime import router as runtime_router
 from .runs import router as runs_router
+from .dynamic_team import router as dynamic_team_router
 
 __all__ = [
     "agents_router",
@@ -20,4 +21,5 @@ __all__ = [
     "guard_router",
     "runtime_router",
     "runs_router",
+    "dynamic_team_router",
 ]

backend/api/dynamic_team.py

@@ -0,0 +1,404 @@
+# -*- coding: utf-8 -*-
+"""Dynamic Team API - REST endpoints for managing analyst team dynamically.
+
+This module provides API endpoints for:
+- Creating new analysts with custom configuration
+- Cloning existing analysts
+- Removing analysts
+- Listing available analyst types
+- Getting analyst information
+- Managing team composition
+
+These endpoints allow both the PM agent (via tool calls) and frontend
+(via HTTP) to manage the analyst team dynamically.
+"""
+from __future__ import annotations
+
+import logging
+from pathlib import Path
+from typing import Any, Dict, List, Optional
+
+from fastapi import APIRouter, HTTPException
+from pydantic import BaseModel, Field
+
+from backend.agents.dynamic_team_types import (
+    AnalystPersona,
+    AnalystConfig,
+    AnalystTypeInfo,
+)
+from backend.config.constants import ANALYST_TYPES
+from backend.agents.prompt_loader import get_prompt_loader
+
+logger = logging.getLogger(__name__)
+
+router = APIRouter(prefix="/api/dynamic-team", tags=["dynamic-team"])
+
+PROJECT_ROOT = Path(__file__).resolve().parents[2]
+
+
+# Pydantic models for API requests/responses
+
+class AnalystPersonaRequest(BaseModel):
+    """Request model for analyst persona definition."""
+    name: str = Field(..., description="Display name for the analyst")
+    focus: List[str] = Field(default_factory=list, description="List of focus areas")
+    description: str = Field(..., description="Detailed description")
+    preferred_tools: Optional[List[str]] = Field(None, description="Preferred tool categories")
+    icon: Optional[str] = Field(None, description="Icon identifier")
+
+
+class CreateAnalystRequest(BaseModel):
+    """Request model for creating a new analyst."""
+    agent_id: str = Field(..., description="Unique identifier for the new analyst")
+    analyst_type: str = Field(..., description="Base type or custom identifier")
+    persona: Optional[AnalystPersonaRequest] = Field(None, description="Custom persona definition")
+    soul_md: Optional[str] = Field(None, description="Custom SOUL.md content")
+    agents_md: Optional[str] = Field(None, description="Custom AGENTS.md content")
+    profile_md: Optional[str] = Field(None, description="Custom PROFILE.md content")
+    bootstrap_md: Optional[str] = Field(None, description="Custom BOOTSTRAP.md content")
+    model_name: Optional[str] = Field(None, description="Override default LLM model")
+    skills: Optional[List[str]] = Field(None, description="List of skill IDs to enable")
+    tags: Optional[List[str]] = Field(None, description="Classification tags")
+
+
+class CloneAnalystRequest(BaseModel):
+    """Request model for cloning an analyst."""
+    source_id: str = Field(..., description="ID of the analyst to clone")
+    new_id: str = Field(..., description="Unique identifier for the new analyst")
+    name: Optional[str] = Field(None, description="New display name")
+    focus_additions: Optional[List[str]] = Field(None, description="Additional focus areas")
+    description_override: Optional[str] = Field(None, description="New description")
+    model_name: Optional[str] = Field(None, description="Override model from source")
+
+
+class RegisterTypeRequest(BaseModel):
+    """Request model for registering a new analyst type."""
+    type_id: str = Field(..., description="Unique identifier for this type")
+    name: str = Field(..., description="Display name")
+    focus: List[str] = Field(..., description="List of focus areas")
+    description: str = Field(..., description="Detailed description")
+    preferred_tools: Optional[List[str]] = Field(None, description="Preferred tool categories")
+
+
+class AnalystResponse(BaseModel):
+    """Response model for analyst operations."""
+    success: bool
+    agent_id: Optional[str] = None
+    message: str
+    error: Optional[str] = None
+
+
+class AnalystTypeResponse(BaseModel):
+    """Response model for analyst type information."""
+    type_id: str
+    name: str
+    description: str
+    is_builtin: bool
+    source: str
+
+
+class AnalystInfoResponse(BaseModel):
+    """Response model for detailed analyst information."""
+    found: bool
+    agent_id: str
+    config: Optional[Dict[str, Any]] = None
+    is_custom: bool = False
+    is_clone: bool = False
+    parent_id: Optional[str] = None
+    message: Optional[str] = None
+
+
+class TeamSummaryResponse(BaseModel):
+    """Response model for team summary."""
+    total_analysts: int
+    custom_analysts: int
+    cloned_analysts: int
+    analysts: List[Dict[str, Any]]
+    registered_types: int
+
+
+# Helper function to get the current pipeline instance
+
+def _get_pipeline(run_id: str) -> Optional[Any]:
+    """Get the TradingPipeline instance for a run.
+
+    Args:
+        run_id: The run configuration ID
+
+    Returns:
+        TradingPipeline instance or None if not found
+    """
+    # Import here to avoid circular imports
+    try:
+        from backend.apps.runtime_service import get_runtime_state
+        runtime_state = get_runtime_state()
+        if runtime_state and hasattr(runtime_state, 'pipeline'):
+            return runtime_state.pipeline
+    except Exception as e:
+        logger.warning(f"Could not get pipeline for run {run_id}: {e}")
+    return None
+
+
+def _get_controller(run_id: str) -> Optional[Any]:
+    """Get the DynamicTeamController for a run.
+
+    Args:
+        run_id: The run configuration ID
+
+    Returns:
+        DynamicTeamController instance or None if not available
+    """
+    try:
+        from backend.tools.dynamic_team_tools import get_controller
+        return get_controller()
+    except Exception as e:
+        logger.warning(f"Could not get controller for run {run_id}: {e}")
+    return None
+
+
+# API Endpoints
+
+@router.get("/types", response_model=List[AnalystTypeResponse])
+async def list_analyst_types() -> List[AnalystTypeResponse]:
+    """List all available analyst types.
+
+    Returns both built-in types (from ANALYST_TYPES) and runtime-registered types.
+    """
+    result = []
+
+    # Add built-in types
+    for type_id, info in ANALYST_TYPES.items():
+        result.append(AnalystTypeResponse(
+            type_id=type_id,
+            name=info.get("display_name", type_id),
+            description=info.get("description", ""),
+            is_builtin=True,
+            source="constants",
+        ))
+
+    # Try to get runtime registered types
+    controller = _get_controller("default")
+    if controller:
+        for type_id, persona in controller._registered_types.items():
+            result.append(AnalystTypeResponse(
+                type_id=type_id,
+                name=persona.name,
+                description=persona.description,
+                is_builtin=False,
+                source="runtime",
+            ))
+
+    return result
+
+
+@router.get("/personas")
+async def get_personas() -> Dict[str, Any]:
+    """Get all analyst personas from personas.yaml.
+
+    Returns the persona definitions used for analyst initialization.
+    """
+    try:
+        personas = get_prompt_loader().load_yaml_config("analyst", "personas")
+        return {"success": True, "personas": personas}
+    except Exception as e:
+        logger.error(f"Failed to load personas: {e}")
+        raise HTTPException(status_code=500, detail=f"Failed to load personas: {e}")
+
+
+@router.post("/runs/{run_id}/analysts", response_model=AnalystResponse)
+async def create_analyst(
+    run_id: str,
+    request: CreateAnalystRequest,
+) -> AnalystResponse:
+    """Create a new analyst in the specified run.
+
+    Args:
+        run_id: The run configuration ID
+        request: Analyst creation configuration
+
+    Returns:
+        Result of the creation operation
+    """
+    controller = _get_controller(run_id)
+    if not controller:
+        raise HTTPException(
+            status_code=503,
+            detail="Dynamic team controller not available. Is the pipeline running?"
+        )
+
+    # Build persona if provided
+    persona = None
+    if request.persona:
+        persona = AnalystPersona(
+            name=request.persona.name,
+            focus=request.persona.focus,
+            description=request.persona.description,
+            preferred_tools=request.persona.preferred_tools,
+            icon=request.persona.icon,
+        )
+
+    # Build config
+    config = AnalystConfig(
+        persona=persona,
+        analyst_type=request.analyst_type if request.analyst_type in ANALYST_TYPES else None,
+        soul_md=request.soul_md,
+        agents_md=request.agents_md,
+        profile_md=request.profile_md,
+        bootstrap_md=request.bootstrap_md,
+        model_name=request.model_name,
+        skills=request.skills or [],
+        tags=request.tags or [],
+    )
+
+    # Create the analyst
+    result = controller.create_analyst(
+        agent_id=request.agent_id,
+        analyst_type=request.analyst_type,
+        name=persona.name if persona else None,
+        focus=persona.focus if persona else None,
+        description=persona.description if persona else None,
+        soul_md=config.soul_md,
+        agents_md=config.agents_md,
+        model_name=config.model_name,
+    )
+
+    return AnalystResponse(**result)
+
+
+@router.post("/runs/{run_id}/analysts/clone", response_model=AnalystResponse)
+async def clone_analyst(
+    run_id: str,
+    request: CloneAnalystRequest,
+) -> AnalystResponse:
+    """Clone an existing analyst.
+
+    Args:
+        run_id: The run configuration ID
+        request: Clone configuration
+
+    Returns:
+        Result of the clone operation
+    """
+    controller = _get_controller(run_id)
+    if not controller:
+        raise HTTPException(
+            status_code=503,
+            detail="Dynamic team controller not available. Is the pipeline running?"
+        )
+
+    result = controller.clone_analyst(
+        source_id=request.source_id,
+        new_id=request.new_id,
+        name=request.name,
+        focus_additions=request.focus_additions,
+        description_override=request.description_override,
+        model_name=request.model_name,
+    )
+
+    return AnalystResponse(**result)
+
+
+@router.delete("/runs/{run_id}/analysts/{agent_id}", response_model=AnalystResponse)
+async def remove_analyst(run_id: str, agent_id: str) -> AnalystResponse:
+    """Remove a dynamically created analyst.
+
+    Args:
+        run_id: The run configuration ID
+        agent_id: The analyst to remove
+
+    Returns:
+        Result of the removal operation
+    """
+    controller = _get_controller(run_id)
+    if not controller:
+        raise HTTPException(
+            status_code=503,
+            detail="Dynamic team controller not available. Is the pipeline running?"
+        )
+
+    result = controller.remove_analyst(agent_id)
+    return AnalystResponse(**result)
+
+
+@router.get("/runs/{run_id}/analysts/{agent_id}", response_model=AnalystInfoResponse)
+async def get_analyst_info(run_id: str, agent_id: str) -> AnalystInfoResponse:
+    """Get information about a specific analyst.
+
+    Args:
+        run_id: The run configuration ID
+        agent_id: The analyst ID
+
+    Returns:
+        Analyst configuration and status
+    """
+    controller = _get_controller(run_id)
+    if not controller:
+        raise HTTPException(
+            status_code=503,
+            detail="Dynamic team controller not available. Is the pipeline running?"
+        )
+
+    result = controller.get_analyst_info(agent_id)
+    return AnalystInfoResponse(**result)
+
+
+@router.get("/runs/{run_id}/summary", response_model=TeamSummaryResponse)
+async def get_team_summary(run_id: str) -> TeamSummaryResponse:
+    """Get a summary of the current analyst team.
+
+    Args:
+        run_id: The run configuration ID
+
+    Returns:
+        Team composition information
+    """
+    controller = _get_controller(run_id)
+    if not controller:
+        raise HTTPException(
+            status_code=503,
+            detail="Dynamic team controller not available. Is the pipeline running?"
+        )
+
+    result = controller.get_team_summary()
+    return TeamSummaryResponse(**result)
+
+
+@router.post("/runs/{run_id}/types", response_model=AnalystTypeResponse)
+async def register_analyst_type(
+    run_id: str,
+    request: RegisterTypeRequest,
+) -> AnalystTypeResponse:
+    """Register a new analyst type.
+
+    Args:
+        run_id: The run configuration ID
+        request: Type registration configuration
+
+    Returns:
+        Registered type information
+    """
+    controller = _get_controller(run_id)
+    if not controller:
+        raise HTTPException(
+            status_code=503,
+            detail="Dynamic team controller not available. Is the pipeline running?"
+        )
+
+    result = controller.register_analyst_type(
+        type_id=request.type_id,
+        name=request.name,
+        focus=request.focus,
+        description=request.description,
+        preferred_tools=request.preferred_tools,
+    )
+
+    if not result.get("success", False):
+        raise HTTPException(status_code=400, detail=result.get("message", "Registration failed"))
+
+    return AnalystTypeResponse(
+        type_id=request.type_id,
+        name=request.name,
+        description=request.description,
+        is_builtin=False,
+        source="runtime",
+    )

backend/apps/runtime_service.py

@@ -5,7 +5,7 @@ from __future__ import annotations
 
 from fastapi import FastAPI
 
-from backend.api import runtime_router
+from backend.api import runtime_router, dynamic_team_router
 from backend.api.runtime import get_runtime_state, _check_gateway_health, _get_gateway_process_details
 from backend.apps.cors import add_cors_middleware
 
@@ -78,6 +78,7 @@ def create_app() -> FastAPI:
         }
 
     app.include_router(runtime_router)
+    app.include_router(dynamic_team_router)
     return app
 
 

backend/config/bootstrap_config.py

@@ -131,6 +131,13 @@ def _coerce_bool(value: Any) -> bool:
     return bool(value)
 
 
+def _normalize_schedule_mode(value: Any) -> str:
+    mode = str(value or "daily").strip().lower()
+    if mode == "intraday":
+        return "interval"
+    return mode or "daily"
+
+
 def resolve_runtime_config(
     project_root: Path,
     config_name: str,
@@ -162,9 +169,9 @@ def resolve_runtime_config(
                 get_env_int("MAX_COMM_CYCLES", 2),
             ),
         ),
-        "schedule_mode": str(
+        "schedule_mode": _normalize_schedule_mode(
             bootstrap.get("schedule_mode", schedule_mode),
-        ).strip().lower() or schedule_mode,
+        ),
         "interval_minutes": int(
             bootstrap.get(
                 "interval_minutes",

backend/core/apo.py

@@ -0,0 +1,197 @@
+# -*- coding: utf-8 -*-
+"""
+Autonomous Policy Optimizer (APO)
+Automatically tunes agent policies based on performance feedback.
+"""
+
+import logging
+import json
+import os
+from datetime import datetime
+from pathlib import Path
+from typing import Any, Dict, List, Optional
+
+from agentscope.message import Msg
+from backend.llm.models import get_agent_model, get_agent_formatter
+from backend.agents.workspace_manager import WorkspaceManager
+
+logger = logging.getLogger(__name__)
+
+class PolicyOptimizer:
+    """
+    PolicyOptimizer analyzes trading performance and automatically updates
+    agent workspace files (POLICY.md, AGENTS.md) to improve future results.
+    """
+
+    def __init__(self, config_name: str, project_root: Optional[Path] = None):
+        self.config_name = config_name
+        self.workspace_manager = WorkspaceManager(project_root=project_root)
+        # Use a high-capability model for the optimizer (meta-agent)
+        self.model = get_agent_model("portfolio_manager") 
+        self.formatter = get_agent_formatter("portfolio_manager")
+
+    async def run_optimization(
+        self,
+        date: str,
+        reflection_content: str,
+        settlement_result: Dict[str, Any],
+        analyst_results: List[Dict[str, Any]],
+        decisions: Dict[str, Dict],
+    ) -> Dict[str, Any]:
+        """
+        Run the optimization loop if performance indicates a need for change.
+        """
+        total_pnl = settlement_result.get("portfolio_value", 0) - 100000.0 # Assuming 100k initial
+        # You might want to use a more sophisticated trigger, like 3 consecutive losses
+        if total_pnl >= 0:
+            logger.info(f"APO: Positive P&L (${total_pnl:,.2f}) for {date}, skipping optimization.")
+            return {"status": "skipped", "reason": "positive_pnl"}
+
+        logger.info(f"APO: Negative P&L (${total_pnl:,.2f}) detected for {date}. Starting optimization...")
+
+        # 1. Identify underperforming agents or logic
+        # 2. Generate policy updates
+        # 3. Apply updates
+        
+        optimizations = []
+        
+        # Focus on agents that gave high confidence but wrong direction
+        underperformers = self._identify_underperformers(settlement_result, analyst_results)
+        
+        for agent_id in underperformers:
+            update = await self._generate_policy_update(
+                agent_id, 
+                date, 
+                reflection_content, 
+                settlement_result,
+                analyst_results,
+                decisions
+            )
+            if update:
+                self._apply_update(agent_id, update)
+                optimizations.append({
+                    "agent_id": agent_id,
+                    "file": update.get("file", "POLICY.md"),
+                    "change": update.get("change", "")
+                })
+
+        return {
+            "status": "completed",
+            "date": date,
+            "total_pnl": total_pnl,
+            "optimizations": optimizations
+        }
+
+    def _identify_underperformers(
+        self, 
+        settlement_result: Dict[str, Any], 
+        analyst_results: List[Dict[str, Any]]
+    ) -> List[str]:
+        """Identify which agents might need policy adjustments."""
+        underperformers = []
+        
+        # Simple logic: if the overall day was a loss, all active analysts might need a check,
+        # but specifically those whose predictions didn't match the market.
+        # For now, let's include all analysts involved in the day.
+        for result in analyst_results:
+            agent_id = result.get("agent")
+            if agent_id:
+                underperformers.append(agent_id)
+        
+        # Also include PM and Risk Manager as they are critical
+        underperformers.append("portfolio_manager")
+        underperformers.append("risk_manager")
+        
+        return list(set(underperformers))
+
+    async def _generate_policy_update(
+        self,
+        agent_id: str,
+        date: str,
+        reflection_content: str,
+        settlement_result: Dict[str, Any],
+        analyst_results: List[Dict[str, Any]],
+        decisions: Dict[str, Dict],
+    ) -> Optional[Dict[str, str]]:
+        """Use LLM to generate a specific policy update for an agent."""
+        
+        # Load current policy
+        try:
+            current_policy = self.workspace_manager.load_agent_file(
+                config_name=self.config_name,
+                agent_id=agent_id,
+                filename="POLICY.md"
+            )
+        except Exception:
+            current_policy = "No existing policy found."
+
+        prompt = f"""
+As an Expert Meta-Optimizer for a multi-agent trading system, your task is to update the operational POLICY for an agent named '{agent_id}' based on recent performance failures.
+
+[Current Context]
+Date: {date}
+Daily Reflection:
+{reflection_content}
+
+[Agent's Current POLICY.md]
+{current_policy}
+
+[Task]
+Analyze why the system failed (loss occurred). Identify what '{agent_id}' could have done differently or what new constraint/heuristic should be added to its policy to prevent similar mistakes in the future.
+
+Provide a specific, concise addition or modification to the POLICY.md file. 
+The output MUST be a JSON object with:
+1. "reasoning": Brief explanation of why this change is needed.
+2. "file": Always "POLICY.md".
+3. "change": The EXACT markdown text to APPEND or REPLACE in the file. Keep it in Chinese as the system uses Chinese prompts.
+
+Output ONLY the JSON object.
+"""
+        msg = Msg(name="system", content=prompt, role="user")
+        response = await self.model.reply(msg)
+        
+        content = response.content
+        if isinstance(content, list):
+            content = content[0].get("text", "")
+            
+        # Clean JSON if wrapped in markdown
+        if "```json" in content:
+            content = content.split("```json")[1].split("```")[0].strip()
+        
+        try:
+            return json.loads(content)
+        except Exception as e:
+            logger.error(f"APO: Failed to parse optimization response for {agent_id}: {e}")
+            return None
+
+    def _apply_update(self, agent_id: str, update: Dict[str, str]) -> None:
+        """Apply the suggested update to the agent's workspace."""
+        filename = update.get("file", "POLICY.md")
+        change = update.get("change", "")
+        
+        if not change:
+            return
+
+        try:
+            current_content = self.workspace_manager.load_agent_file(
+                config_name=self.config_name,
+                agent_id=agent_id,
+                filename=filename
+            )
+            
+            # Check if change is already there to avoid duplicates
+            if change.strip() in current_content:
+                logger.info(f"APO: Change already present in {agent_id}/{filename}")
+                return
+
+            new_content = current_content + "\n\n### APO Update (" + datetime.now().strftime("%Y-%m-%d") + ")\n" + change
+            
+            self.workspace_manager.update_agent_file(
+                config_name=self.config_name,
+                agent_id=agent_id,
+                filename=filename,
+                content=new_content
+            )
+            logger.info(f"APO: Updated {agent_id}/{filename} with new heuristics.")
+        except Exception as e:
+            logger.error(f"APO: Failed to apply update to {agent_id}/{filename}: {e}")

backend/core/pipeline.py

@@ -33,6 +33,8 @@ from backend.agents.workspace_manager import WorkspaceManager
 from backend.agents.prompt_loader import get_prompt_loader
 from backend.llm.models import get_agent_formatter, get_agent_model
 from backend.config.constants import ANALYST_TYPES
+from backend.agents.dynamic_team_types import AnalystConfig
+from backend.tools.dynamic_team_tools import DynamicTeamController, set_controller
 
 
 def _resolve_evo_agent_ids() -> set[str]:
@@ -84,6 +86,9 @@ def _log(msg: str) -> None:
     logger.info(msg)
 
 
+from backend.core.apo import PolicyOptimizer
+
+
 class TradingPipeline:
     """
     Trading Pipeline - Orchestrates the complete trading cycle
@@ -127,7 +132,21 @@ class TradingPipeline:
         self.runtime_manager = runtime_manager
         self._session_key: Optional[str] = None
         self._dynamic_analysts: Dict[str, Any] = {}
+        self._dynamic_analyst_configs: Dict[str, AnalystConfig] = {}
 
+        # Initialize APO (Autonomous Policy Optimizer)
+        config_name = workspace_id or (runtime_manager.config_name if runtime_manager else "default")
+        self.apo = PolicyOptimizer(config_name=config_name)
+
+        # Initialize dynamic team controller and inject into PM
+        self._team_controller = DynamicTeamController(
+            create_callback=self._create_runtime_analyst,
+            remove_callback=self._remove_runtime_analyst,
+            get_analysts_callback=self._all_analysts,
+        )
+        set_controller(self._team_controller)
+
+        # Backward compatibility: also set individual callbacks if PM expects them
         if hasattr(self.pm, "set_team_controller"):
             self.pm.set_team_controller(
                 create_agent_callback=self._create_runtime_analyst,
@@ -150,23 +169,7 @@ class TradingPipeline:
         execute_decisions: bool = True,
     ) -> Dict[str, Any]:
         """
-        Run one complete trading cycle
-
-        Args:
-            tickers: List of stock tickers
-            date: Trading date (YYYY-MM-DD)
-            prices: Open prices {ticker: price} (for backtest)
-            close_prices: Close prices for settlement (for backtest)
-            market_caps: Optional market caps for baseline calculation
-            get_open_prices_fn: Async callback to wait for open prices (live mode)
-            get_close_prices_fn: Async callback to wait for close prices (live mode)
-
-        For live mode:
-        - Analysis runs immediately
-        - Execution waits for market open via get_open_prices_fn
-        - Settlement waits for market close via get_close_prices_fn
-
-        Each agent's result is broadcast immediately via StateSync.
+        Run one complete trading cycle with checkpointing support.
         """
         _log(f"Starting cycle {date} - {len(tickers)} tickers")
         session_key = TradingSessionKey(date=date).key()
@@ -176,12 +179,43 @@ class TradingPipeline:
             agents=active_analysts + [self.risk_manager, self.pm],
             session_key=session_key,
         )
+
+        # Load checkpoint if exists
+        checkpoint = self._load_checkpoint(session_key)
+        checkpoint_data = checkpoint.get("data", {}) if checkpoint else {}
+        last_phase = checkpoint.get("phase") if checkpoint else None
+
+        if checkpoint:
+            _log(f"Resuming from checkpoint: {last_phase}")
+            # Restore state from checkpoint
+            analyst_results = checkpoint_data.get("analyst_results", [])
+            risk_assessment = checkpoint_data.get("risk_assessment", {})
+            self.conference_summary = checkpoint_data.get("conference_summary")
+            final_predictions = checkpoint_data.get("final_predictions", [])
+            pm_result = checkpoint_data.get("pm_result", {})
+            execution_result = checkpoint_data.get("execution_result", {})
+            settlement_result = checkpoint_data.get("settlement_result")
+            # Prefer passed prices if not hold in checkpoint
+            if not prices:
+                prices = checkpoint_data.get("prices")
+            if not close_prices:
+                close_prices = checkpoint_data.get("close_prices")
+        else:
+            analyst_results = []
+            risk_assessment = {}
+            self.conference_summary = None
+            final_predictions = []
+            pm_result = {}
+            execution_result = {}
+            settlement_result = None
+
         if self.runtime_manager:
             self.runtime_manager.set_session_key(session_key)
-            self._runtime_log_event("cycle:start", {"tickers": tickers, "date": date})
+            self._runtime_log_event("cycle:start", {"tickers": tickers, "date": date, "resumed": checkpoint is not None})
             self._runtime_batch_status(active_analysts, "analysis_in_progress")
 
-        # Phase 0: Clear short-term memory to avoid cross-day context pollution
+        # Phase 0: Clear memory (only if not resuming or if resuming from very start)
+        if not last_phase:
             _log("Phase 0: Clearing memory")
             await self._clear_all_agent_memory()
 
@@ -196,15 +230,23 @@ class TradingPipeline:
                 "system",
             ),
         ):
-            # Phase 1.1: Analysts (parallel execution with TeamCoordinator)
+            # Phase 1.1: Analysts
+            if not last_phase or last_phase == "cleared":
                 _log("Phase 1.1: Analyst analysis (parallel)")
                 analyst_results = await self._run_analysts_parallel(
                     tickers,
                     date,
                     active_analysts=active_analysts,
                 )
+                self._save_checkpoint(session_key, "analysis", {
+                    "analyst_results": analyst_results,
+                    "prices": prices,
+                    "close_prices": close_prices
+                })
+                last_phase = "analysis"
 
             # Phase 1.2: Risk Manager
+            if last_phase == "analysis":
                 _log("Phase 1.2: Risk assessment")
                 self._runtime_update_status(self.risk_manager, "risk_assessment")
                 risk_assessment = await self._run_risk_manager_with_sync(
@@ -212,8 +254,16 @@ class TradingPipeline:
                     date,
                     prices,
                 )
+                self._save_checkpoint(session_key, "risk_assessment", {
+                    "analyst_results": analyst_results,
+                    "risk_assessment": risk_assessment,
+                    "prices": prices,
+                    "close_prices": close_prices
+                })
+                last_phase = "risk_assessment"
 
-            # Phase 2.1: Conference discussion (within same MsgHub)
+            # Phase 2.1: Conference discussion
+            if last_phase == "risk_assessment":
                 _log("Phase 2.1: Conference discussion")
                 conference_summary = await self._run_conference_cycles(
                     tickers=tickers,
@@ -223,28 +273,56 @@ class TradingPipeline:
                     risk_assessment=risk_assessment,
                 )
                 self.conference_summary = conference_summary
+                self._save_checkpoint(session_key, "conference", {
+                    "analyst_results": analyst_results,
+                    "risk_assessment": risk_assessment,
+                    "conference_summary": conference_summary,
+                    "prices": prices,
+                    "close_prices": close_prices
+                })
+                last_phase = "conference"
 
             # Phase 2.2: Analysts generate final structured predictions
+            if last_phase == "conference":
                 _log("Phase 2.2: Analysts generate final structured predictions")
                 final_predictions = await self._collect_final_predictions(
                     tickers,
                     date,
                     active_analysts=active_analysts,
                 )
+                self._save_checkpoint(session_key, "predictions", {
+                    "analyst_results": analyst_results,
+                    "risk_assessment": risk_assessment,
+                    "conference_summary": conference_summary,
+                    "final_predictions": final_predictions,
+                    "prices": prices,
+                    "close_prices": close_prices
+                })
+                last_phase = "predictions"
 
-            # Record final predictions for leaderboard ranking
-            if self.settlement_coordinator:
+            # Record final predictions
+            if last_phase == "predictions" and self.settlement_coordinator:
                 self.settlement_coordinator.record_analyst_predictions(
                     final_predictions,
                 )
 
-            # Live mode: wait for market open before execution
-            if get_open_prices_fn:
+            # Live mode: wait for market open
+            if not prices and get_open_prices_fn:
                 _log("Waiting for market open...")
                 prices = await get_open_prices_fn()
                 _log(f"Got open prices: {prices}")
+                # Update prices in checkpoint if we just got them
+                self._save_checkpoint(session_key, "predictions", {
+                    "analyst_results": analyst_results,
+                    "risk_assessment": risk_assessment,
+                    "conference_summary": conference_summary,
+                    "final_predictions": final_predictions,
+                    "prices": prices,
+                    "close_prices": close_prices
+                })
 
             # Phase 3: PM makes decisions
+            if last_phase == "predictions":
                 _log("Phase 3.1: PM makes decisions")
                 self._runtime_update_status(self.pm, "decision_phase")
                 pm_result = await self._run_pm_with_sync(
@@ -254,12 +332,26 @@ class TradingPipeline:
                     analyst_results,
                     risk_assessment,
                 )
+                self._save_checkpoint(session_key, "decisions", {
+                    "analyst_results": analyst_results,
+                    "risk_assessment": risk_assessment,
+                    "conference_summary": conference_summary,
+                    "final_predictions": final_predictions,
+                    "pm_result": pm_result,
+                    "prices": prices,
+                    "close_prices": close_prices
+                })
+                last_phase = "decisions"
 
-        decisions = pm_result.get("decisions", {})
+        # Outside MsgHub for execution and settlement
+        decisions = pm_result.get("decisions", {}) if pm_result else {}
+        if not execution_result:
             execution_result = {
                 "executed_trades": [],
                 "portfolio": self.pm.get_portfolio_state(),
             }
+
+        if last_phase == "decisions":
             if execute_decisions:
                 _log("Phase 4: Executing trades")
                 self._runtime_update_status(self.pm, "executing")
@@ -267,14 +359,37 @@ class TradingPipeline:
             else:
                 _log("Phase 4: Skipping trade execution")
             
-        # Live mode: wait for market close before settlement
-        if get_close_prices_fn:
+            self._save_checkpoint(session_key, "execution", {
+                "analyst_results": analyst_results,
+                "risk_assessment": risk_assessment,
+                "conference_summary": conference_summary,
+                "final_predictions": final_predictions,
+                "pm_result": pm_result,
+                "execution_result": execution_result,
+                "prices": prices,
+                "close_prices": close_prices
+            })
+            last_phase = "execution"
+
+        # Live mode: wait for market close
+        if not close_prices and get_close_prices_fn:
             _log("Waiting for market close")
             close_prices = await get_close_prices_fn()
             _log(f"Got close prices: {close_prices}")
+            # Update close_prices in checkpoint
+            self._save_checkpoint(session_key, "execution", {
+                "analyst_results": analyst_results,
+                "risk_assessment": risk_assessment,
+                "conference_summary": conference_summary,
+                "final_predictions": final_predictions,
+                "pm_result": pm_result,
+                "execution_result": execution_result,
+                "prices": prices,
+                "close_prices": close_prices
+            })
 
-        # Phase 5: Settlement - run after close prices available
-        settlement_result = None
+        # Phase 5: Settlement
+        if last_phase == "execution":
             if close_prices and self.settlement_coordinator:
                 _log("Phase 5: Daily review and generate memories")
                 self._runtime_batch_status(
@@ -316,6 +431,19 @@ class TradingPipeline:
                     "reflection",
                 )
             
+            self._save_checkpoint(session_key, "settlement", {
+                "analyst_results": analyst_results,
+                "risk_assessment": risk_assessment,
+                "conference_summary": conference_summary,
+                "final_predictions": final_predictions,
+                "pm_result": pm_result,
+                "execution_result": execution_result,
+                "settlement_result": settlement_result,
+                "prices": prices,
+                "close_prices": close_prices
+            })
+            last_phase = "settlement"
+
         _log(f"Cycle complete: {date}")
         self._runtime_batch_status(
             self._all_analysts() + [self.risk_manager, self.pm],
@@ -323,6 +451,11 @@ class TradingPipeline:
         )
         self._runtime_log_event("cycle:end", {"tickers": tickers, "date": date})
 
+        # Optional: Clean up checkpoint after successful completion
+        # path = self._get_checkpoint_path(session_key)
+        # if path and path.exists():
+        #     path.unlink()
+
         return {
             "analyst_results": analyst_results,
             "risk_assessment": risk_assessment,
@@ -385,6 +518,44 @@ class TradingPipeline:
         await self.risk_manager.memory.clear()
         await self.pm.memory.clear()
 
+    def _get_checkpoint_path(self, session_key: str) -> Optional[Path]:
+        """Get the path to the pipeline checkpoint file."""
+        if not self.runtime_manager or not self.runtime_manager.run_dir:
+            return None
+        checkpoint_dir = self.runtime_manager.run_dir / "state" / "checkpoints"
+        checkpoint_dir.mkdir(parents=True, exist_ok=True)
+        return checkpoint_dir / f"pipeline_{session_key}.json"
+
+    def _save_checkpoint(self, session_key: str, phase: str, data: Dict[str, Any]) -> None:
+        """Save the current pipeline state to a checkpoint file."""
+        path = self._get_checkpoint_path(session_key)
+        if not path:
+            return
+
+        checkpoint = {
+            "session_key": session_key,
+            "phase": phase,
+            "timestamp": datetime.now().isoformat(),
+            "data": data
+        }
+        try:
+            path.write_text(json.dumps(checkpoint, ensure_ascii=False, indent=2, default=str), encoding="utf-8")
+            _log(f"Checkpoint saved: {phase} for {session_key}")
+        except Exception as e:
+            logger.error(f"Failed to save checkpoint: {e}")
+
+    def _load_checkpoint(self, session_key: str) -> Optional[Dict[str, Any]]:
+        """Load the pipeline state from a checkpoint file."""
+        path = self._get_checkpoint_path(session_key)
+        if not path or not path.exists():
+            return None
+
+        try:
+            return json.loads(path.read_text(encoding="utf-8"))
+        except Exception as e:
+            logger.error(f"Failed to load checkpoint: {e}")
+            return None
+
     async def _sync_memory_if_retrieved(self, agent: Any) -> None:
         """
         Check agent's short-term memory for retrieved long-term memory and sync to frontend.
@@ -585,6 +756,25 @@ class TradingPipeline:
                 content=reflection_content,
             )
 
+        # Phase 6: APO (Autonomous Policy Optimization)
+        # If the day was a loss, let APO suggest and apply policy updates.
+        if hasattr(self, "apo") and self.apo:
+            _log(f"Phase 6: APO - Running autonomous policy optimization for {date}")
+            try:
+                apo_result = await self.apo.run_optimization(
+                    date=date,
+                    reflection_content=reflection_content,
+                    settlement_result=settlement_result or {"portfolio_value": 100000.0 + total_pnl},
+                    analyst_results=analyst_results,
+                    decisions=decisions
+                )
+                if apo_result.get("status") == "completed":
+                    _log(f"APO: Successfully applied {len(apo_result.get('optimizations', []))} policy updates.")
+                    # Reload assets for next cycle to ensure they are picked up
+                    self.reload_runtime_assets()
+            except Exception as e:
+                logger.error(f"APO: Optimization failed: {e}")
+
     def _build_reflection_content(
         self,
         date: str,
@@ -1562,28 +1752,74 @@ class TradingPipeline:
         """Return static analysts plus runtime-created analysts."""
         return list(self.analysts) + list(self._dynamic_analysts.values())
 
-    def _create_runtime_analyst(self, agent_id: str, analyst_type: str) -> str:
-        """Create one runtime analyst instance."""
-        if analyst_type not in ANALYST_TYPES:
+    def _create_runtime_analyst(
+        self,
+        agent_id: str,
+        analyst_type: str,
+        custom_config: Optional[AnalystConfig] = None,
+    ) -> str:
+        """Create one runtime analyst instance.
+
+        Args:
+            agent_id: Unique identifier for the new analyst
+            analyst_type: Type of analyst (e.g., "technical_analyst")
+            custom_config: Optional custom configuration for the analyst,
+                          including persona, soul_md, agents_md, etc.
+
+        Returns:
+            Success or error message
+        """
+        # Validate analyst_type or custom_config
+        if analyst_type not in ANALYST_TYPES and not custom_config:
             return (
                 f"Unknown analyst_type '{analyst_type}'. "
-                f"Available: {', '.join(ANALYST_TYPES.keys())}"
+                f"Available: {', '.join(ANALYST_TYPES.keys())}. "
+                f"Or provide custom_config to create a custom analyst."
             )
         if agent_id in {agent.name for agent in self._all_analysts()}:
             return f"Analyst '{agent_id}' already exists."
 
         config_name = getattr(self.pm, "config", {}).get("config_name", "default")
         project_root = Path(__file__).resolve().parents[2]
+
+        # Get persona: use custom_config if provided, else load from personas.yaml
+        if custom_config and custom_config.persona:
+            persona = {
+                "name": custom_config.persona.name,
+                "focus": custom_config.persona.focus,
+                "description": custom_config.persona.description,
+            }
+        else:
             personas = get_prompt_loader().load_yaml_config("analyst", "personas")
             persona = personas.get(analyst_type, {})
         workspace_manager = WorkspaceManager(project_root=project_root)
+
+        # Build file contents: use custom if provided, else generate from persona
+        file_contents = {}
+        if custom_config:
+            if custom_config.soul_md:
+                file_contents["SOUL.md"] = custom_config.soul_md
+            if custom_config.agents_md:
+                file_contents["AGENTS.md"] = custom_config.agents_md
+            if custom_config.profile_md:
+                file_contents["PROFILE.md"] = custom_config.profile_md
+            if custom_config.bootstrap_md:
+                file_contents["BOOTSTRAP.md"] = custom_config.bootstrap_md
+
+        # Fill in any missing files with defaults
+        if not file_contents or len(file_contents) < 4:
+            default_files = workspace_manager.build_default_agent_files(
+                agent_id=agent_id,
+                persona=persona,
+            )
+            for key, value in default_files.items():
+                if key not in file_contents:
+                    file_contents[key] = value
+
         workspace_manager.ensure_agent_assets(
             config_name=config_name,
             agent_id=agent_id,
-            file_contents=workspace_manager.build_default_agent_files(
-                agent_id=agent_id,
-                persona=persona,
-            ),
+            file_contents=file_contents,
         )
 
         # Create EvoAgent with workspace-driven configuration
@@ -1594,11 +1830,23 @@ class TradingPipeline:
             agent_id,
         )
         agent_config = load_agent_workspace_config(workspace_dir / "agent.yaml")
+        # Support model override from custom_config
+        if custom_config and custom_config.model_name:
+            # Import create_model for custom model creation
+            from backend.llm.models import create_model
+            # Use specified model name, default to openai provider
+            model = create_model(
+                model_name=custom_config.model_name,
+                model_provider=custom_config.memory_config.get("model_provider", "openai") if custom_config.memory_config else "openai"
+            )
+        else:
+            model = get_agent_model(analyst_type)
+
         agent = EvoAgent(
             agent_id=agent_id,
             config_name=config_name,
             workspace_dir=workspace_dir,
-            model=get_agent_model(analyst_type),
+            model=model,
             formatter=get_agent_formatter(analyst_type),
             prompt_files=agent_config.prompt_files,
         )
@@ -1611,6 +1859,11 @@ class TradingPipeline:
         # Keep workspace_id for backward compatibility
         setattr(agent, "workspace_id", config_name)
         self._dynamic_analysts[agent_id] = agent
+
+        # Store custom config for future reference (e.g., cloning)
+        if custom_config:
+            self._dynamic_analyst_configs[agent_id] = custom_config
+
         update_active_analysts(
             project_root=project_root,
             config_name=config_name,
@@ -1624,6 +1877,8 @@ class TradingPipeline:
         if agent_id not in self._dynamic_analysts:
             return f"Runtime analyst '{agent_id}' not found."
         self._dynamic_analysts.pop(agent_id, None)
+        # Also remove stored config if exists
+        self._dynamic_analyst_configs.pop(agent_id, None)
         config_name = getattr(self.pm, "config", {}).get("config_name", "default")
         project_root = Path(__file__).resolve().parents[2]
         update_active_analysts(

backend/core/scheduler.py

@@ -17,6 +17,14 @@ NYSE_TZ = ZoneInfo("America/New_York")
 NYSE_CALENDAR = mcal.get_calendar("NYSE")
 
 
+def normalize_schedule_mode(mode: str | None) -> str:
+    """Normalize schedule mode to the current public vocabulary."""
+    value = str(mode or "daily").strip().lower()
+    if value == "intraday":
+        return "interval"
+    return value or "daily"
+
+
 class Scheduler:
     """
     Market-aware scheduler for live trading.
@@ -31,7 +39,7 @@ class Scheduler:
         heartbeat_interval: Optional[int] = None,
         config: Optional[dict] = None,
     ):
-        self.mode = mode
+        self.mode = normalize_schedule_mode(mode)
         self.trigger_time = trigger_time or "09:30"  # NYSE timezone
         self.trigger_now = self.trigger_time == "now"
         self.interval_minutes = interval_minutes or 60
@@ -107,7 +115,7 @@ class Scheduler:
 
         if self.mode == "daily":
             self._task = asyncio.create_task(self._run_daily(self._callback))
-        elif self.mode == "intraday":
+        elif self.mode == "interval":
             self._task = asyncio.create_task(
                 self._run_intraday(self._callback),
             )
@@ -124,8 +132,13 @@ class Scheduler:
         """Update scheduler parameters in-place and restart its timing loop."""
         changed = False
 
-        if mode and mode != self.mode:
-            self.mode = mode
+        if mode:
+            normalized_mode = normalize_schedule_mode(mode)
+        else:
+            normalized_mode = None
+
+        if normalized_mode and normalized_mode != self.mode:
+            self.mode = normalized_mode
             changed = True
 
         if trigger_time and trigger_time != self.trigger_time:
@@ -233,13 +246,13 @@ class Scheduler:
             await callback(date=current_date)
 
     async def _run_intraday(self, callback: Callable):
-        """Run every N minutes (for future use)"""
+        """Run every N minutes in interval mode."""
         while self.running:
             now = self._now_nyse()
             current_date = now.strftime("%Y-%m-%d")
 
             if self._is_trading_day(now):
-                logger.info(f"Triggering intraday cycle for {current_date}")
+                logger.info(f"Triggering interval cycle for {current_date}")
                 await callback(date=current_date)
 
             await asyncio.sleep(self.interval_minutes * 60)

backend/services/gateway_admin_handlers.py

@@ -25,6 +25,13 @@ from backend.config.bootstrap_config import (
 from backend.llm.models import get_agent_model_info
 
 
+def _normalize_schedule_mode(value: Any) -> str:
+    mode = str(value or "daily").strip().lower()
+    if mode == "intraday":
+        return "interval"
+    return mode or "daily"
+
+
 async def handle_reload_runtime_assets(gateway: Any) -> None:
     config_name = gateway.config.get("config_name", "default")
     runtime_config = resolve_runtime_config(
@@ -44,10 +51,10 @@ async def handle_reload_runtime_assets(gateway: Any) -> None:
 async def handle_update_runtime_config(gateway: Any, websocket: Any, data: dict[str, Any]) -> None:
     updates: dict[str, Any] = {}
 
-    schedule_mode = str(data.get("schedule_mode", "")).strip().lower()
+    schedule_mode = _normalize_schedule_mode(data.get("schedule_mode", ""))
     if schedule_mode:
-        if schedule_mode not in {"daily", "intraday"}:
-            await websocket.send(json.dumps({"type": "error", "message": "schedule_mode must be 'daily' or 'intraday'."}, ensure_ascii=False))
+        if schedule_mode not in {"daily", "interval"}:
+            await websocket.send(json.dumps({"type": "error", "message": "schedule_mode must be 'daily' or 'interval'."}, ensure_ascii=False))
             return
         updates["schedule_mode"] = schedule_mode
 

backend/services/gateway_cycle_support.py

@@ -208,7 +208,7 @@ async def run_live_cycle(gateway: Any, date: str, tickers: list[str]) -> None:
     market_status = gateway.market_service.get_market_status()
     current_prices = gateway.market_service.get_all_prices()
 
-    if schedule_mode == "intraday":
+    if schedule_mode in {"interval", "intraday"}:
         execute_decisions = market_status.get("status") == "open"
         if execute_decisions:
             await gateway.state_sync.on_system_message("定时任务触发：当前处于交易时段，本轮将执行交易决策")

backend/services/gateway_runtime_support.py

@@ -8,6 +8,13 @@ from typing import Any
 from backend.data.provider_utils import normalize_symbol
 
 
+def _normalize_schedule_mode(value: Any) -> str:
+    mode = str(value or "daily").strip().lower()
+    if mode == "intraday":
+        return "interval"
+    return mode or "daily"
+
+
 def normalize_watchlist(raw_tickers: Any) -> list[str]:
     """Parse watchlist payloads from websocket messages."""
     if raw_tickers is None:
@@ -51,9 +58,11 @@ def apply_runtime_config(gateway: Any, runtime_config: dict[str, Any]) -> dict[s
 
     gateway.pipeline.max_comm_cycles = int(runtime_config["max_comm_cycles"])
     gateway.config["max_comm_cycles"] = gateway.pipeline.max_comm_cycles
-    gateway.config["schedule_mode"] = runtime_config.get(
+    gateway.config["schedule_mode"] = _normalize_schedule_mode(
+        runtime_config.get(
             "schedule_mode",
             gateway.config.get("schedule_mode", "daily"),
+        ),
     )
     gateway.config["interval_minutes"] = int(
         runtime_config.get(

backend/skills/builtin/dynamic_team_management/SKILL.md

@@ -0,0 +1,189 @@
+---
+name: dynamic_team_management
+description: 动态管理团队中的分析师Agent，包括创建、克隆、移除分析师，以及查看可用分析师类型。
+version: 1.0.0
+tools:
+  - create_analyst
+  - clone_analyst
+  - remove_analyst
+  - list_analyst_types
+  - get_analyst_info
+  - get_team_summary
+---
+
+# 动态团队管理
+
+当你需要调整分析师团队组成时，使用这个技能。投资经理可以动态创建新的分析师、克隆现有分析师进行定制、或移除不再需要的分析师。
+
+## 1) When to use
+
+- 当前团队缺乏特定领域的分析能力（如期权、加密货币、ESG等）
+- 需要多个相同类型但不同配置的分析师（如"激进型技术分析师"和"保守型技术分析师"）
+- 需要临时增加分析力量应对特殊市场环境
+- 发现某个分析师配置不当，需要移除并重建
+- 在团队讨论中发现需要新的分析视角
+
+## 2) Required inputs
+
+### 创建分析师 (create_analyst)
+- **agent_id**: 唯一标识符（如 "options_specialist_01"）
+- **analyst_type**: 基础类型（如 "technical_analyst"）或自定义标识
+- **可选**: name, focus, description, soul_md, agents_md, model_name
+
+### 克隆分析师 (clone_analyst)
+- **source_id**: 源分析师ID（如 "technical_analyst"）
+- **new_id**: 新分析师ID（如 "crypto_technical_01"）
+- **可选**: name, focus_additions, description_override, model_name
+
+### 移除分析师 (remove_analyst)
+- **agent_id**: 要移除的分析师ID
+
+## 3) Decision procedure
+
+1. **评估当前团队能力缺口**
+   - 查看当前活跃的分析师列表
+   - 识别缺失的分析视角或专业领域
+
+2. **选择创建策略**
+   - 基于现有类型创建：指定analyst_type，提供自定义配置
+   - 完全自定义：提供完整的persona定义
+   - 克隆并修改：从现有分析师复制并应用覆盖
+
+3. **配置分析师**
+   - 设置唯一agent_id
+   - 定义显示名称和关注点
+   - 可选：提供自定义SOUL.md内容以精确定义行为
+
+4. **验证创建结果**
+   - 检查返回的success状态
+   - 确认新分析师已加入活跃列表
+
+## 4) Tool call policy
+
+- **create_analyst**: 用于创建全新的分析师实例
+  - 必须提供唯一的agent_id
+  - 基于预定义类型时，analyst_type必须在可用类型列表中，或提供完整自定义配置
+  - 工具调用失败时，检查agent_id是否已存在
+
+- **clone_analyst**: 用于基于现有分析师创建变体
+  - 适用于：创建专注于特定行业的分析师（如从technical_analyst克隆crypto_technical）
+  - 新实例继承源配置，应用指定的覆盖
+
+- **remove_analyst**: 用于移除动态创建的分析师
+  - 只能移除通过本技能创建的分析师
+  - 系统预定义分析师（fundamentals_analyst等）不可移除
+
+- **list_analyst_types**: 用于查看所有可用分析师类型
+  - 返回预定义类型 + 运行时注册类型
+
+- **get_analyst_info**: 用于查看特定分析师的详细配置
+
+- **get_team_summary**: 用于查看团队整体构成
+
+## 5) Output schema
+
+### create_analyst / clone_analyst 输出
+```json
+{
+  "success": true,
+  "agent_id": "options_specialist_01",
+  "message": "Created runtime analyst 'options_specialist_01' (technical_analyst).",
+  "analyst_type": "technical_analyst"
+}
+```
+
+### remove_analyst 输出
+```json
+{
+  "success": true,
+  "agent_id": "options_specialist_01",
+  "message": "Removed runtime analyst 'options_specialist_01'."
+}
+```
+
+### list_analyst_types 输出
+```json
+[
+  {
+    "type_id": "fundamentals_analyst",
+    "name": "Fundamentals Analyst",
+    "description": "...",
+    "is_builtin": true,
+    "source": "constants"
+  }
+]
+```
+
+## 6) Failure fallback
+
+- **agent_id已存在**: 返回错误，提示选择新的agent_id或使用clone_analyst基于现有创建变体
+- **analyst_type未知**: 提示使用list_analyst_types查看可用类型，或提供完整的自定义persona
+- **创建失败**: 检查系统日志，可能原因包括：模型配置错误、工作空间权限问题
+- **移除失败**: 确认分析师是通过动态创建（系统预定义分析师不可移除）
+
+## 重要约定
+
+### Agent ID 命名规则
+
+为了使新创建的分析师能够正常工作，**agent_id 必须以 `_analyst` 结尾**。这是系统识别分析师类型并分配相应工具的关键约定。
+
+- ✅ **正确**: `options_specialist_analyst`, `crypto_technical_analyst`
+- ❌ **错误**: `options_specialist`, `crypto_expert`
+
+如果不遵循此约定，分析师将无法获得分析工具组（基本面、技术、情绪、估值等工具）。
+
+### 全新自定义类型 vs 基于现有类型
+
+**基于现有类型**（推荐用于快速创建）：
+- 使用 `analyst_type: "technical_analyst"` 等预定义类型
+- 可以覆盖 persona、SOUL.md 等配置
+- 工具组根据 `analyst_type` 自动选择
+
+**全新自定义类型**（用于完全自定义）：
+- 设置 `analyst_type` 为自定义标识（如 `"custom"`）或任意字符串
+- 必须提供完整的 `persona` 定义
+- 建议提供 `soul_md` 精确定义行为
+- **agent_id 必须仍然以 `_analyst` 结尾**
+
+## 最佳实践
+
+1. **命名约定**: 使用描述性agent_id，如 `industry_tech_analyst` 而非 `analyst_01`，**必须以 `_analyst` 结尾**
+2. **版本控制**: 克隆分析师时，在new_id中包含版本信息，如 `technical_v2_crypto_analyst`
+3. **文档记录**: 创建自定义分析师时，提供详细的description，便于后续理解和维护
+4. **资源管理**: 定期使用get_team_summary检查团队规模，移除不再需要的分析师
+
+## 示例场景
+
+### 场景1: 添加加密货币分析师
+```
+创建一个新的分析师，专注于加密货币技术分析：
+- agent_id: "crypto_technical_01"
+- analyst_type: "technical_analyst"
+- name: "加密货币技术分析师"
+- focus: ["链上数据分析", "DeFi协议", "加密货币技术指标"]
+```
+
+### 场景2: 克隆并定制
+```
+基于technical_analyst创建一个更激进的版本：
+- source_id: "technical_analyst"
+- new_id: "technical_aggressive_01"
+- name: "激进技术分析师"
+- focus_additions: ["高波动交易", "突破策略"]
+- description_override: "专注于高风险高回报的技术策略..."
+```
+
+### 场景3: 创建全新自定义类型（期权专家）
+```
+创建一个完全自定义的期权分析师（注意agent_id以_analyst结尾）：
+- agent_id: "options_strategist_analyst"
+- analyst_type: "custom"  # 使用非预定义类型
+- name: "期权策略分析师"
+- focus: ["期权定价", "希腊字母", "波动率曲面"]
+- soul_md: "# 角色定义\n你是期权策略专家，专注于..."
+```
+
+**说明**: 
+- 即使 `analyst_type` 是 "custom"（不在预定义类型中），只要提供完整的 `persona` 和 `soul_md`，系统就能创建功能完整的分析师
+- `agent_id` 必须以 `_analyst` 结尾才能获得分析工具
+- 模型使用全局默认，或通过 `model_name` 参数指定

backend/tests/test_gateway_support_modules.py

@@ -159,11 +159,11 @@ def test_apply_runtime_config_updates_gateway_state():
     )
 
     assert gateway.config["tickers"] == ["MSFT", "NVDA"]
-    assert gateway.config["schedule_mode"] == "intraday"
+    assert gateway.config["schedule_mode"] == "interval"
     assert gateway.storage.initial_cash == 150000.0
     assert result["runtime_config_applied"]["max_comm_cycles"] == 4
     assert gateway.scheduler.calls[-1] == {
-        "mode": "intraday",
+        "mode": "interval",
         "trigger_time": "10:30",
         "interval_minutes": 30,
     }

backend/tests/test_runtime_service_app.py

@@ -86,7 +86,7 @@ def test_runtime_service_get_runtime_config(monkeypatch, tmp_path):
         "---\n"
         "tickers:\n"
         "  - AAPL\n"
-        "schedule_mode: intraday\n"
+        "schedule_mode: interval\n"
         "interval_minutes: 30\n"
         "trigger_time: '10:00'\n"
         "max_comm_cycles: 3\n"
@@ -102,7 +102,7 @@ def test_runtime_service_get_runtime_config(monkeypatch, tmp_path):
                     "run_dir": str(run_dir),
                     "bootstrap_values": {
                         "tickers": ["AAPL"],
-                        "schedule_mode": "intraday",
+                        "schedule_mode": "interval",
                         "interval_minutes": 30,
                         "trigger_time": "10:00",
                         "max_comm_cycles": 3,
@@ -123,7 +123,7 @@ def test_runtime_service_get_runtime_config(monkeypatch, tmp_path):
     assert response.status_code == 200
     payload = response.json()
     assert payload["run_id"] == "demo"
-    assert payload["bootstrap"]["schedule_mode"] == "intraday"
+    assert payload["bootstrap"]["schedule_mode"] == "interval"
     assert payload["resolved"]["interval_minutes"] == 30
     assert payload["resolved"]["enable_memory"] is True
 
@@ -190,7 +190,7 @@ def test_runtime_service_update_runtime_config_persists_bootstrap(monkeypatch, t
         response = client.put(
             "/api/runtime/config",
             json={
-                "schedule_mode": "intraday",
+                "schedule_mode": "interval",
                 "interval_minutes": 15,
                 "trigger_time": "10:15",
                 "max_comm_cycles": 4,
@@ -199,7 +199,7 @@ def test_runtime_service_update_runtime_config_persists_bootstrap(monkeypatch, t
 
     assert response.status_code == 200
     payload = response.json()
-    assert payload["bootstrap"]["schedule_mode"] == "intraday"
+    assert payload["bootstrap"]["schedule_mode"] == "interval"
     assert payload["resolved"]["interval_minutes"] == 15
     assert "interval_minutes: 15" in (run_dir / "BOOTSTRAP.md").read_text(encoding="utf-8")
 
@@ -547,7 +547,7 @@ def test_start_runtime_restore_reuses_historical_run_id(monkeypatch, tmp_path):
                     "run_dir": str(run_dir),
                     "bootstrap_values": {
                         "tickers": ["AAPL"],
-                        "schedule_mode": "intraday",
+                        "schedule_mode": "interval",
                         "interval_minutes": 30,
                         "trigger_time": "now",
                         "max_comm_cycles": 2,

backend/tools/dynamic_team_tools.py

@@ -0,0 +1,518 @@
+# -*- coding: utf-8 -*-
+"""Dynamic Team Management Tools - Tools for PM to manage analyst team dynamically.
+
+This module provides tools for the Portfolio Manager to:
+- Create new analysts with custom configuration
+- Clone existing analysts with variations
+- Remove analysts from the team
+- List available analyst types
+- Get analyst information
+
+These tools are registered with the PM's toolkit and enable dynamic team management
+as described in the Dynamic Team Architecture.
+"""
+from __future__ import annotations
+
+from typing import Any, Dict, List, Optional, Callable
+from dataclasses import asdict
+
+from backend.agents.dynamic_team_types import (
+    AnalystPersona,
+    AnalystConfig,
+    CreateAnalystResult,
+    AnalystTypeInfo,
+)
+from backend.config.constants import ANALYST_TYPES
+
+
+# Type alias for callbacks set by pipeline
+CreateAnalystCallback = Callable[[str, str, Optional[AnalystConfig]], str]
+RemoveAnalystCallback = Callable[[str], str]
+
+
+class DynamicTeamController:
+    """Controller for dynamic analyst team management.
+
+    This class is instantiated by TradingPipeline and injected into the PM agent
+    via set_team_controller(). It provides methods that the PM can call through
+    tools to manage the analyst team dynamically.
+
+    Attributes:
+        create_callback: Callback to _create_runtime_analyst in pipeline
+        remove_callback: Callback to _remove_runtime_analyst in pipeline
+        get_analysts_callback: Callback to get current analysts list
+        registered_types: Runtime-registered custom analyst types
+    """
+
+    def __init__(
+        self,
+        create_callback: CreateAnalystCallback,
+        remove_callback: RemoveAnalystCallback,
+        get_analysts_callback: Optional[Callable[[], List[Any]]] = None,
+    ):
+        """Initialize the controller with callbacks from pipeline.
+
+        Args:
+            create_callback: Function to create a runtime analyst
+            remove_callback: Function to remove a runtime analyst
+            get_analysts_callback: Optional function to get current analysts
+        """
+        self._create_callback = create_callback
+        self._remove_callback = remove_callback
+        self._get_analysts_callback = get_analysts_callback
+        self._registered_types: Dict[str, AnalystPersona] = {}
+        self._instance_configs: Dict[str, AnalystConfig] = {}
+
+    def create_analyst(
+        self,
+        agent_id: str,
+        analyst_type: str,
+        name: Optional[str] = None,
+        focus: Optional[List[str]] = None,
+        description: Optional[str] = None,
+        soul_md: Optional[str] = None,
+        agents_md: Optional[str] = None,
+        model_name: Optional[str] = None,
+        preferred_tools: Optional[List[str]] = None,
+    ) -> Dict[str, Any]:
+        """Create a new analyst with optional custom configuration.
+
+        This tool allows the Portfolio Manager to dynamically create new analysts
+        during a trading session. The analyst can be based on a predefined type
+        or fully customized with a unique persona.
+
+        Args:
+            agent_id: Unique identifier for the new analyst (e.g., "crypto_specialist_01")
+            analyst_type: Base type (e.g., "technical_analyst") or custom identifier
+            name: Display name for the analyst (overrides default)
+            focus: List of focus areas (overrides default)
+            description: Detailed description (overrides default)
+            soul_md: Custom SOUL.md content for the analyst's workspace
+            agents_md: Custom AGENTS.md content
+            model_name: Override the default LLM model
+            preferred_tools: List of preferred tool categories
+
+        Returns:
+            Dict with success status, message, and analyst info
+
+        Example:
+            >>> result = create_analyst(
+            ...     agent_id="options_specialist",
+            ...     analyst_type="technical_analyst",
+            ...     name="期权策略分析师",
+            ...     focus=["期权定价", "波动率交易"],
+            ...     description="专注于期权市场分析和波动率交易策略...",
+            ... )
+        """
+        # Build custom config if any customization is provided
+        custom_config = None
+        if any([name, focus, description, soul_md, agents_md, model_name, preferred_tools]):
+            persona = None
+            if name or focus or description:
+                persona = AnalystPersona(
+                    name=name or f"Custom {analyst_type}",
+                    focus=focus or ["General Analysis"],
+                    description=description or f"Custom analyst based on {analyst_type}",
+                    preferred_tools=preferred_tools,
+                )
+
+            custom_config = AnalystConfig(
+                persona=persona,
+                analyst_type=analyst_type if analyst_type in ANALYST_TYPES else None,
+                soul_md=soul_md,
+                agents_md=agents_md,
+                model_name=model_name,
+            )
+
+        # Call the pipeline's create method
+        result_message = self._create_callback(agent_id, analyst_type, custom_config)
+
+        # Parse result
+        success = result_message.startswith("Created")
+        if success:
+            self._instance_configs[agent_id] = custom_config if custom_config else AnalystConfig(
+                analyst_type=analyst_type
+            )
+
+        return {
+            "success": success,
+            "agent_id": agent_id if success else None,
+            "message": result_message,
+            "analyst_type": analyst_type,
+        }
+
+    def clone_analyst(
+        self,
+        source_id: str,
+        new_id: str,
+        name: Optional[str] = None,
+        focus_additions: Optional[List[str]] = None,
+        description_override: Optional[str] = None,
+        model_name: Optional[str] = None,
+    ) -> Dict[str, Any]:
+        """Clone an existing analyst with optional modifications.
+
+        Creates a new analyst by copying the configuration of an existing one
+        and applying specified overrides. Useful for creating specialized
+        variants (e.g., "crypto_technical" from "technical_analyst").
+
+        Args:
+            source_id: ID of the analyst to clone
+            new_id: Unique identifier for the new analyst
+            name: New display name (if different from source)
+            focus_additions: Additional focus areas to add
+            description_override: Completely new description
+            model_name: Override the model from source
+
+        Returns:
+            Dict with success status, message, and new analyst info
+
+        Example:
+            >>> result = clone_analyst(
+            ...     source_id="technical_analyst",
+            ...     new_id="crypto_technical_01",
+            ...     name="加密货币技术分析师",
+            ...     focus_additions=["链上数据", "DeFi协议分析"],
+            ... )
+        """
+        # Get source config if available
+        source_config = self._instance_configs.get(source_id)
+
+        # Determine base type and config
+        if source_config:
+            base_type = source_config.analyst_type or source_id
+            base_persona = source_config.persona
+        else:
+            # Assume source_id is a known type
+            base_type = source_id
+            base_persona = None
+
+        # Build new persona
+        new_focus = list(base_persona.focus) if base_persona else []
+        if focus_additions:
+            new_focus.extend(focus_additions)
+
+        new_name = name or (base_persona.name if base_persona else new_id)
+        new_description = description_override or (base_persona.description if base_persona else "")
+
+        # Create new config with parent reference
+        new_config = AnalystConfig(
+            persona=AnalystPersona(
+                name=new_name,
+                focus=new_focus,
+                description=new_description,
+                preferred_tools=base_persona.preferred_tools if base_persona else None,
+            ),
+            analyst_type=base_type if base_type in ANALYST_TYPES else None,
+            soul_md=source_config.soul_md if source_config else None,
+            agents_md=source_config.agents_md if source_config else None,
+            model_name=model_name or (source_config.model_name if source_config else None),
+            parent_id=source_id,
+        )
+
+        # Create the new analyst
+        result_message = self._create_callback(new_id, base_type, new_config)
+
+        success = result_message.startswith("Created")
+        if success:
+            self._instance_configs[new_id] = new_config
+
+        return {
+            "success": success,
+            "agent_id": new_id if success else None,
+            "parent_id": source_id,
+            "message": result_message,
+        }
+
+    def remove_analyst(self, agent_id: str) -> Dict[str, Any]:
+        """Remove a dynamically created analyst from the team.
+
+        Args:
+            agent_id: ID of the analyst to remove
+
+        Returns:
+            Dict with success status and message
+
+        Example:
+            >>> result = remove_analyst("options_specialist")
+        """
+        result_message = self._remove_callback(agent_id)
+        success = result_message.startswith("Removed") or "not found" not in result_message.lower()
+
+        if success and agent_id in self._instance_configs:
+            del self._instance_configs[agent_id]
+
+        return {
+            "success": success,
+            "agent_id": agent_id,
+            "message": result_message,
+        }
+
+    def list_analyst_types(self) -> List[Dict[str, Any]]:
+        """List all available analyst types.
+
+        Returns a list of all available analyst types, including:
+        - Built-in types from ANALYST_TYPES
+        - Runtime registered custom types
+
+        Returns:
+            List of analyst type information dictionaries
+
+        Example:
+            >>> types = list_analyst_types()
+            >>> print(types[0]["type_id"])  # "fundamentals_analyst"
+        """
+        result = []
+
+        # Add built-in types
+        for type_id, info in ANALYST_TYPES.items():
+            result.append({
+                "type_id": type_id,
+                "name": info.get("display_name", type_id),
+                "description": info.get("description", ""),
+                "is_builtin": True,
+                "source": "constants",
+            })
+
+        # Add runtime registered types
+        for type_id, persona in self._registered_types.items():
+            result.append({
+                "type_id": type_id,
+                "name": persona.name,
+                "description": persona.description,
+                "is_builtin": False,
+                "source": "runtime",
+            })
+
+        return result
+
+    def get_analyst_info(self, agent_id: str) -> Dict[str, Any]:
+        """Get information about a specific analyst.
+
+        Args:
+            agent_id: ID of the analyst
+
+        Returns:
+            Dict with analyst configuration and status
+        """
+        config = self._instance_configs.get(agent_id)
+        if not config:
+            return {
+                "found": False,
+                "agent_id": agent_id,
+                "message": f"No configuration found for '{agent_id}'",
+            }
+
+        return {
+            "found": True,
+            "agent_id": agent_id,
+            "config": config.to_dict(),
+            "is_custom": config.persona is not None,
+            "is_clone": config.parent_id is not None,
+            "parent_id": config.parent_id,
+        }
+
+    def register_analyst_type(
+        self,
+        type_id: str,
+        name: str,
+        focus: List[str],
+        description: str,
+        preferred_tools: Optional[List[str]] = None,
+    ) -> Dict[str, Any]:
+        """Register a new analyst type for later creation.
+
+        This allows defining reusable analyst personas that can be instantiated
+        multiple times with different configurations.
+
+        Args:
+            type_id: Unique identifier for this type (e.g., "options_analyst")
+            name: Display name
+            focus: List of focus areas
+            description: Detailed description
+            preferred_tools: Optional list of preferred tool categories
+
+        Returns:
+            Dict with success status and type info
+
+        Example:
+            >>> result = register_analyst_type(
+            ...     type_id="options_analyst",
+            ...     name="期权分析师",
+            ...     focus=["期权定价", "希腊字母分析"],
+            ...     description="专注于期权策略和波动率分析",
+            ... )
+        """
+        if type_id in self._registered_types or type_id in ANALYST_TYPES:
+            return {
+                "success": False,
+                "type_id": type_id,
+                "message": f"Type '{type_id}' already exists",
+            }
+
+        persona = AnalystPersona(
+            name=name,
+            focus=focus,
+            description=description,
+            preferred_tools=preferred_tools,
+        )
+        self._registered_types[type_id] = persona
+
+        return {
+            "success": True,
+            "type_id": type_id,
+            "persona": persona.to_dict(),
+            "message": f"Registered new analyst type '{type_id}'",
+        }
+
+    def get_team_summary(self) -> Dict[str, Any]:
+        """Get a summary of the current analyst team.
+
+        Returns:
+            Dict with team composition information
+        """
+        analysts = []
+        for agent_id, config in self._instance_configs.items():
+            analysts.append({
+                "agent_id": agent_id,
+                "name": config.persona.name if config.persona else agent_id,
+                "type": config.analyst_type,
+                "is_custom": config.persona is not None,
+                "is_clone": config.parent_id is not None,
+            })
+
+        return {
+            "total_analysts": len(analysts),
+            "custom_analysts": len([a for a in analysts if a["is_custom"]]),
+            "cloned_analysts": len([a for a in analysts if a["is_clone"]]),
+            "analysts": analysts,
+            "registered_types": len(self._registered_types),
+        }
+
+
+# Global controller instance - set by pipeline
+_controller_instance: Optional[DynamicTeamController] = None
+
+
+def set_controller(controller: DynamicTeamController) -> None:
+    """Set the global controller instance.
+
+    Called by TradingPipeline when initializing the PM agent.
+    """
+    global _controller_instance
+    _controller_instance = controller
+
+
+def get_controller() -> Optional[DynamicTeamController]:
+    """Get the global controller instance.
+
+    Returns:
+        DynamicTeamController instance or None if not set
+    """
+    return _controller_instance
+
+
+# Tool functions that wrap the controller methods
+# These are registered with the PM's toolkit
+
+def create_analyst(
+    agent_id: str,
+    analyst_type: str,
+    name: Optional[str] = None,
+    focus: Optional[str] = None,
+    description: Optional[str] = None,
+    soul_md: Optional[str] = None,
+    agents_md: Optional[str] = None,
+    model_name: Optional[str] = None,
+) -> Dict[str, Any]:
+    """Tool wrapper for create_analyst.
+
+    Note: focus parameter accepts comma-separated string for tool compatibility.
+    """
+    controller = get_controller()
+    if not controller:
+        return {"success": False, "error": "Dynamic team controller not available"}
+
+    focus_list = [f.strip() for f in focus.split(",")] if focus else None
+    return controller.create_analyst(
+        agent_id=agent_id,
+        analyst_type=analyst_type,
+        name=name,
+        focus=focus_list,
+        description=description,
+        soul_md=soul_md,
+        agents_md=agents_md,
+        model_name=model_name,
+    )
+
+
+def clone_analyst(
+    source_id: str,
+    new_id: str,
+    name: Optional[str] = None,
+    focus_additions: Optional[str] = None,
+    description_override: Optional[str] = None,
+    model_name: Optional[str] = None,
+) -> Dict[str, Any]:
+    """Tool wrapper for clone_analyst.
+
+    Note: focus_additions accepts comma-separated string.
+    """
+    controller = get_controller()
+    if not controller:
+        return {"success": False, "error": "Dynamic team controller not available"}
+
+    additions_list = [f.strip() for f in focus_additions.split(",")] if focus_additions else None
+    return controller.clone_analyst(
+        source_id=source_id,
+        new_id=new_id,
+        name=name,
+        focus_additions=additions_list,
+        description_override=description_override,
+        model_name=model_name,
+    )
+
+
+def remove_analyst(agent_id: str) -> Dict[str, Any]:
+    """Tool wrapper for remove_analyst."""
+    controller = get_controller()
+    if not controller:
+        return {"success": False, "error": "Dynamic team controller not available"}
+    return controller.remove_analyst(agent_id)
+
+
+def list_analyst_types() -> List[Dict[str, Any]]:
+    """Tool wrapper for list_analyst_types."""
+    controller = get_controller()
+    if not controller:
+        return []
+    return controller.list_analyst_types()
+
+
+def get_analyst_info(agent_id: str) -> Dict[str, Any]:
+    """Tool wrapper for get_analyst_info."""
+    controller = get_controller()
+    if not controller:
+        return {"found": False, "error": "Controller not available"}
+    return controller.get_analyst_info(agent_id)
+
+
+def get_team_summary() -> Dict[str, Any]:
+    """Tool wrapper for get_team_summary."""
+    controller = get_controller()
+    if not controller:
+        return {"error": "Controller not available"}
+    return controller.get_team_summary()
+
+
+__all__ = [
+    "DynamicTeamController",
+    "set_controller",
+    "get_controller",
+    "create_analyst",
+    "clone_analyst",
+    "remove_analyst",
+    "list_analyst_types",
+    "get_analyst_info",
+    "get_team_summary",
+]