feat: add official Codex skill and harden task execution

Install and update artemis-use with task diagnostics and plugin guidance. Preserve cross-device file receipts, review canonical reports, guard stage transitions and reduce small history summaries. Improve exploration lifecycle, request recovery and browser live connections.

tests/unit/test_exploration_report_output.py

@@ -0,0 +1,39 @@
+"""The default SDK path must return the answer already prepared for review."""
+
+import json
+from types import SimpleNamespace
+from unittest.mock import AsyncMock
+
+import pytest
+
+from artemis.config import OutputConfig
+from artemis.graph.state import State
+from artemis.sdk.agent import Agent
+
+
+@pytest.mark.asyncio
+@pytest.mark.parametrize("outputter_enabled", [True, False])
+async def test_default_output_reuses_executor_report_without_another_model(
+    tmp_path, monkeypatch, outputter_enabled
+):
+    synthesize = AsyncMock()
+    monkeypatch.setattr("artemis.sdk.agent.outputter", synthesize)
+    ctx = SimpleNamespace(
+        execution_setup=SimpleNamespace(
+            outputter=SimpleNamespace(enabled=outputter_enabled, force_synthesis=False),
+            disable_outputter=not outputter_enabled,
+        )
+    )
+    result_path = tmp_path / "result.json"
+    body = "# 调查结果\n双端检查已完成；未测功能不能记为通过。"
+    result = await Agent._extract_output(
+        object.__new__(Agent),
+        task_name="exploration",
+        ctx=ctx,
+        request=SimpleNamespace(llm_output_path=result_path),
+        output_config=OutputConfig(structured_output=None, output_description=None),
+        state=State(initial_goal="调查", execution_report=body),
+    )
+    assert result == body
+    assert json.loads(result_path.read_text(encoding="utf-8")) == {"result": body}
+    synthesize.assert_not_called()