Ship Cursor-only MCP admin and modular Admin/Chat UI.

Plugins/Skills install and edit via mcpServers JSON, with clearer row actions, soft reloads, and instant loading placeholders; also drop MCP market and harden related scheduler/logging/exec guards.

Co-authored-by: Cursor <cursoragent@cursor.com>

tests/test_scheduled_job_failure_board.py

@@ -0,0 +1,73 @@
+# -*- coding: utf-8 -*-
+"""Tests for scheduled job failure classification + summary."""
+
+from __future__ import annotations
+
+from pathlib import Path
+
+from runtime.scheduler.failure_class import classify_scheduled_job_error, enrich_scheduled_job_run_dict
+from svc.persistence.sqlite_store import SqliteStore
+
+
+def test_classify_scheduled_job_error_classes() -> None:
+    assert classify_scheduled_job_error("overlapping_run", status="skipped") == "overlap"
+    assert classify_scheduled_job_error("read_timeout after 60s", status="failed") == "timeout"
+    assert classify_scheduled_job_error("WhatsApp Connection Closed", status="failed") == "delivery"
+    assert classify_scheduled_job_error("insufficient_scope:sql:query", status="failed") == "auth"
+    assert classify_scheduled_job_error("Unregistered tool: mcp__netx__x", status="failed") == "mcp"
+    assert classify_scheduled_job_error("stale_running_cleared", status="failed") == "stale"
+    assert classify_scheduled_job_error("boom", status="failed") == "runtime"
+    assert classify_scheduled_job_error("", status="success") == ""
+
+
+def test_enrich_run_dict_adds_failure_class() -> None:
+    out = enrich_scheduled_job_run_dict({"status": "failed", "error": "tool timeout"})
+    assert out["failure_class"] == "timeout"
+
+
+def test_scheduled_job_failure_summary(tmp_path: Path) -> None:
+    store = SqliteStore(str(tmp_path / "jobs.sqlite"))
+    # Ensure tenant user exists if required by store create paths.
+    job = store.scheduled_job_create(
+        tenant_id="default",
+        name="fail-board",
+        description="",
+        schedule_kind="interval",
+        schedule_expr="3600",
+        timezone_name="UTC",
+        prompt_text="ping",
+        specialist="ops",
+        lang="en",
+        created_by_user_id="u1",
+        source="test",
+    )
+    run_ok = store.scheduled_job_run_create(
+        job_id=job.id, tenant_id="default", scheduled_at="2026-08-01T00:00:00+00:00", status="queued"
+    )
+    store.scheduled_job_run_update(
+        run_id=run_ok.id,
+        tenant_id="default",
+        patch={"status": "success", "finished_at": "2026-08-01T00:01:00+00:00", "error": ""},
+    )
+    run_fail = store.scheduled_job_run_create(
+        job_id=job.id, tenant_id="default", scheduled_at="2026-08-01T01:00:00+00:00", status="queued"
+    )
+    store.scheduled_job_run_update(
+        run_id=run_fail.id,
+        tenant_id="default",
+        patch={
+            "status": "failed",
+            "finished_at": "2026-08-01T01:01:00+00:00",
+            "error": "read_timeout after 90s",
+        },
+    )
+    store.scheduled_job_mark_run(
+        job_id=job.id, tenant_id="default", last_run_status="failed", pause_after=False
+    )
+    summary = store.scheduled_job_failure_summary(tenant_id="default", recent_limit=50)
+    assert summary["recent_failed"] >= 1
+    assert summary["recent_success"] >= 1
+    assert "timeout" in (summary.get("recent_fail_classes") or {})
+    assert any(x.get("id") == job.id for x in summary.get("jobs_last_failed") or [])
+    d = store.scheduled_job_run_to_dict(store.scheduled_job_run_get(run_id=run_fail.id, tenant_id="default"))
+    assert d.get("failure_class") == "timeout"