feat: add agent visibility APIs and harden runtime verification

Add Day 4 visibility endpoints and response models, strengthen collaboration/task verification behavior, and patch conversation schema startup migration for agent_state compatibility. Extend backend regression coverage for runtime schemas, verifier behavior, visibility APIs, router auth, and legacy conversation list loading.

backend/app/agents/schemas/__init__.py

@@ -1,10 +1,25 @@
 from app.agents.schemas.event import AgentEvent
-from app.agents.schemas.task import AgentTask, TaskResult, TaskLifecycleStatus, VerificationStatus
+from app.agents.schemas.message import AgentMessage
+from app.agents.schemas.task import (
+    AgentTask,
+    CollaborationBudget,
+    InterruptRecord,
+    RecoveryRecord,
+    TaskLifecycleStatus,
+    TaskResult,
+    TaskResultStatus,
+    VerificationStatus,
+)
 
 __all__ = [
     "AgentEvent",
+    "AgentMessage",
     "AgentTask",
+    "CollaborationBudget",
+    "InterruptRecord",
+    "RecoveryRecord",
     "TaskLifecycleStatus",
     "TaskResult",
+    "TaskResultStatus",
     "VerificationStatus",
 ]

backend/app/agents/schemas/event.py

@@ -11,6 +11,18 @@ AgentEventType = Literal[
     "agent.tool.result",
     "agent.verify.started",
     "agent.verify.completed",
+    "agent.created",
+    "agent.spawn.blocked",
+    "agent.message.sent",
+    "agent.message.received",
+    "agent.interrupt.requested",
+    "agent.interrupt.completed",
+    "agent.recovery.started",
+    "agent.recovery.completed",
+    "agent.task.interrupted",
+    "agent.task.recovered",
+    "agent.task.reassigned",
+    "agent.collaboration.budget.updated",
     "agent.error",
 ]
 AgentEventSeverity = Literal["info", "warning", "error"]
@@ -24,5 +36,11 @@ class AgentEvent(BaseModel):
     agent_id: str | None = None
     sub_commander_id: str | None = None
     task_id: str | None = None
+    parent_task_id: str | None = None
+    child_task_id: str | None = None
+    thread_id: str | None = None
+    message_id: str | None = None
+    interrupt_id: str | None = None
+    recovery_id: str | None = None
     payload: dict[str, Any] = Field(default_factory=dict)
     severity: AgentEventSeverity = "info"

backend/app/agents/schemas/message.py

@@ -0,0 +1,29 @@
+from __future__ import annotations
+
+from datetime import datetime, timezone
+from typing import Any, Literal
+
+from pydantic import BaseModel, Field
+
+
+AgentMessageType = Literal[
+    "task_request",
+    "task_update",
+    "handoff",
+    "verification_request",
+    "verification_feedback",
+    "interrupt_notice",
+]
+
+
+class AgentMessage(BaseModel):
+    message_id: str
+    thread_id: str
+    from_agent_id: str
+    to_agent_id: str
+    task_id: str | None = None
+    reply_to_message_id: str | None = None
+    message_type: AgentMessageType = "task_update"
+    content_summary: str
+    created_at: datetime = Field(default_factory=lambda: datetime.now(timezone.utc))
+    payload: dict[str, Any] = Field(default_factory=dict)

backend/app/agents/schemas/task.py

@@ -8,6 +8,41 @@ from pydantic import BaseModel, Field
 
 TaskLifecycleStatus = Literal["pending", "in_progress", "completed", "failed", "blocked"]
 VerificationStatus = Literal["passed", "failed", "skipped"]
+TaskResultStatus = Literal["completed", "failed", "blocked", "passed", "skipped"]
+InterruptStatus = Literal["requested", "acknowledged", "resolved"]
+BudgetMode = Literal["direct", "collaboration"]
+
+
+class InterruptRecord(BaseModel):
+    interrupt_id: str
+    reason: str
+    status: InterruptStatus = "requested"
+    requested_by: str | None = None
+    source_event_id: str | None = None
+    requested_at: datetime = Field(default_factory=lambda: datetime.now(timezone.utc))
+    payload: dict[str, Any] = Field(default_factory=dict)
+
+
+class RecoveryRecord(BaseModel):
+    recovery_id: str
+    source_interrupt_id: str | None = None
+    strategy: str | None = None
+    resumed_from_task_id: str | None = None
+    resumed_from_thread_id: str | None = None
+    recovered_at: datetime = Field(default_factory=lambda: datetime.now(timezone.utc))
+    payload: dict[str, Any] = Field(default_factory=dict)
+
+
+class CollaborationBudget(BaseModel):
+    mode: BudgetMode = "direct"
+    max_parallel_tasks: int | None = None
+    remaining_parallel_tasks: int | None = None
+    max_tool_calls: int | None = None
+    remaining_tool_calls: int | None = None
+    max_iterations: int | None = None
+    remaining_iterations: int | None = None
+    escalation_threshold: int | None = None
+    metadata: dict[str, Any] = Field(default_factory=dict)
 
 
 class AgentTask(BaseModel):
@@ -17,8 +52,16 @@ class AgentTask(BaseModel):
     owner_agent_id: str | None = None
     role: str | None = None
     goal: str | None = None
+    parent_task_id: str | None = None
+    child_task_ids: list[str] = Field(default_factory=list)
+    thread_id: str | None = None
+    message_id: str | None = None
+    message_index: int | None = None
     expected_evidence: list[dict[str, Any]] = Field(default_factory=list)
     evidence: list[dict[str, Any]] = Field(default_factory=list)
+    interrupt_records: list[InterruptRecord | dict[str, Any]] = Field(default_factory=list)
+    recovery_records: list[RecoveryRecord | dict[str, Any]] = Field(default_factory=list)
+    collaboration_budget: CollaborationBudget | dict[str, Any] | None = None
     result_summary: str | None = None
     created_at: datetime = Field(default_factory=lambda: datetime.now(timezone.utc))
     updated_at: datetime = Field(default_factory=lambda: datetime.now(timezone.utc))
@@ -26,7 +69,17 @@ class AgentTask(BaseModel):
 
 class TaskResult(BaseModel):
     task_id: str
-    status: VerificationStatus
+    status: TaskResultStatus
     summary: str | None = None
     evidence: list[dict[str, Any]] = Field(default_factory=list)
+    owner_agent_id: str | None = None
+    parent_task_id: str | None = None
+    child_task_ids: list[str] = Field(default_factory=list)
+    thread_id: str | None = None
+    message_id: str | None = None
+    message_index: int | None = None
+    interrupt_records: list[InterruptRecord | dict[str, Any]] = Field(default_factory=list)
+    recovery_records: list[RecoveryRecord | dict[str, Any]] = Field(default_factory=list)
+    budget_snapshot: CollaborationBudget | dict[str, Any] | None = None
+    next_action: str | None = None
     output_data: dict[str, Any] | None = None