重构主控编排与运行时预热链路，统一工作区提示词/专家调度协议并补齐 wiki 记忆注入与写回闭环。

同时收敛启动与运维脚本默认行为（含 wiki worker）、更新 Admin 可观测性与相关测试，降低首轮时延并提高运行稳定性。

Made-with: Cursor

tests/test_skill_executor_trace.py

@@ -2,6 +2,7 @@ from __future__ import annotations
 
 from types import SimpleNamespace
 
+from oclaw.runtime import skill_executor as skill_executor_mod
 from oclaw.runtime.skill_executor import SkillExecutionContext, SkillExecutor
 from oclaw.runtime.tools.base import ToolRegistry, ToolSpec
 
@@ -47,3 +48,34 @@ def test_skill_executor_emits_skill_trace_events() -> None:
     assert "skill_selected" in et
     assert "skill_executed" in et
 
+
+def test_skill_executor_emits_internal_hook_events(monkeypatch) -> None:
+    calls: list[tuple[str, str]] = []
+
+    def _cap(**kwargs):  # noqa: ANN003
+        t = str(kwargs.get("event_type") or "")
+        a = str(kwargs.get("action") or "")
+        calls.append((t, a))
+
+    monkeypatch.setattr(skill_executor_mod, "trigger_hook_event", _cap)
+    store = _DummyStore()
+    reg = ToolRegistry(
+        [
+            ToolSpec(
+                name="echo_skill",
+                description="echo",
+                parameters={"type": "object", "properties": {"x": {"type": "string"}}},
+                handler=lambda a: {"ok": True, "x": a.get("x")},
+                read_only=True,
+            )
+        ]
+    )
+    ex = SkillExecutor()
+    uses = [SimpleNamespace(id="call_1", name="echo_skill", arguments={"x": "1"})]
+    ex.execute_skill_uses(
+        ctx=SkillExecutionContext(store=store, tools=reg, session_id="s1", trace_id="t1", workspace_dir="/tmp/ws"),
+        assistant_msg_id=1,
+        skill_uses=uses,
+    )
+    assert calls == [("skill", "before"), ("skill", "after")]
+