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/verifier.py

@@ -1,10 +1,10 @@
 from __future__ import annotations
 
-from typing import Any
+from typing import Any, cast
 
 from pydantic import BaseModel, Field
 
-from app.agents.schemas.task import AgentTask, TaskResult, VerificationStatus
+from app.agents.schemas.task import AgentTask, TaskResult, TaskResultStatus, VerificationStatus
 from app.agents.state import AgentState
 
 
@@ -14,6 +14,34 @@ class VerificationVerdict(BaseModel):
     evidence: list[dict[str, Any]] = Field(default_factory=list)
 
 
+def normalize_task_result(
+    task_result: TaskResult | dict[str, Any],
+    *,
+    default_task_id: str | None = None,
+) -> TaskResult:
+    payload = task_result.model_dump(mode="json") if isinstance(task_result, TaskResult) else dict(task_result or {})
+    normalized_status = payload.get("status")
+    if normalized_status not in {"completed", "failed", "blocked", "passed", "skipped"}:
+        normalized_status = "failed"
+    return TaskResult(
+        task_id=str(payload.get("task_id") or default_task_id or "unknown-task"),
+        status=cast(TaskResultStatus, normalized_status),
+        summary=payload.get("summary"),
+        evidence=list(payload.get("evidence") or []),
+        owner_agent_id=payload.get("owner_agent_id"),
+        parent_task_id=payload.get("parent_task_id"),
+        child_task_ids=list(payload.get("child_task_ids") or []),
+        thread_id=payload.get("thread_id"),
+        message_id=payload.get("message_id"),
+        message_index=payload.get("message_index") if isinstance(payload.get("message_index"), int) else None,
+        interrupt_records=list(payload.get("interrupt_records") or []),
+        recovery_records=list(payload.get("recovery_records") or []),
+        budget_snapshot=payload.get("budget_snapshot") if isinstance(payload.get("budget_snapshot"), dict) else None,
+        next_action=payload.get("next_action"),
+        output_data=payload.get("output_data") if isinstance(payload.get("output_data"), dict) else None,
+    )
+
+
 def verify_task_result(
     *,
     task: AgentTask | dict[str, Any] | None = None,
@@ -30,8 +58,13 @@ def verify_task_result(
     if status is not None:
         return VerificationVerdict(status=status, summary=normalized_summary, evidence=normalized_evidence)
 
-    if normalized_result.get("status") in {"passed", "failed", "skipped"}:
-        inferred_status = normalized_result["status"]
+    normalized_status = normalized_result.get("status")
+    if normalized_status in {"passed", "failed", "skipped"}:
+        inferred_status = normalized_status
+    elif normalized_status == "completed":
+        inferred_status = "passed"
+    elif normalized_status == "blocked":
+        inferred_status = "skipped"
     elif normalized_result.get("success") is True:
         inferred_status = "passed"
     elif normalized_result.get("success") is False:
@@ -57,4 +90,4 @@ def apply_verification_verdict(state: AgentState, verdict: VerificationVerdict)
     return AgentState(**next_state)
 
 
-__all__ = ["VerificationVerdict", "apply_verification_verdict", "verify_task_result"]
+__all__ = ["VerificationVerdict", "apply_verification_verdict", "normalize_task_result", "verify_task_result"]