fix(continue): snapshot the active log on reset

frontends/continue_cmd.py

@@ -448,10 +448,10 @@ def _current_log_path(pid=None):
     return os.path.join(_LOG_DIR, f'model_responses_{pid}.txt')
 
 
-def _snapshot_current_log(pid=None):
-    """Persist current PID log as a standalone recoverable snapshot, then clear it."""
-    path = _current_log_path(pid)
-    if not os.path.isfile(path):
+def _snapshot_current_log(pid=None, path=None):
+    """Persist the active log as a standalone recoverable snapshot, then clear it."""
+    path = _current_log_path(pid) if path is None else path
+    if not path or not os.path.isfile(path):
         return None
     try:
         with open(path, encoding='utf-8', errors='replace') as fh:
@@ -477,7 +477,7 @@ def reset_conversation(agent, message='🆕 已开启新对话，当前上下文
         agent.abort()
     except Exception:
         pass
-    _snapshot_current_log()
+    _snapshot_current_log(path=getattr(agent, 'log_path', None))
     if hasattr(agent, 'history'):
         agent.history = []
     for client in _agent_clients(agent):

frontends/tests/test_continue_log_snapshot.py

@@ -0,0 +1,65 @@
+import os
+from importlib.util import module_from_spec, spec_from_file_location
+from pathlib import Path
+from types import SimpleNamespace
+
+
+ROOT = Path(__file__).resolve().parents[2]
+SPEC = spec_from_file_location("continue_cmd_under_test", ROOT / "frontends" / "continue_cmd.py")
+assert SPEC and SPEC.loader
+continue_cmd = module_from_spec(SPEC)
+SPEC.loader.exec_module(continue_cmd)
+
+LOG_CONTENT = (
+    "=== Prompt === 2026-01-01 00:00:00\nhello\n\n"
+    "=== Response === 2026-01-01 00:00:01 model=test\nworld\n\n"
+)
+
+
+def _agent(log_path):
+    backend = SimpleNamespace(history=["backend turn"])
+    client = SimpleNamespace(backend=backend, last_tools="cached tools")
+    return SimpleNamespace(
+        abort=lambda: None, log_path=log_path, history=["agent turn"],
+        llmclients=[client], llmclient=client, handler=object(),
+    )
+
+
+def test_reset_snapshots_the_agents_active_log(tmp_path, monkeypatch):
+    active_log = tmp_path / f"model_responses_{os.getpid() + 1}.txt"
+    active_log.write_text(LOG_CONTENT, encoding="utf-8")
+    agent = _agent(str(active_log))
+    monkeypatch.setattr(continue_cmd, "_LOG_DIR", str(tmp_path))
+
+    continue_cmd.reset_conversation(agent)
+
+    snapshots = list(tmp_path.glob("model_responses_snapshot_*.txt"))
+    assert active_log.read_text(encoding="utf-8") == ""
+    assert len(snapshots) == 1
+    assert snapshots[0].read_text(encoding="utf-8") == LOG_CONTENT
+    assert agent.history == []
+    assert agent.llmclient.backend.history == []
+    assert agent.llmclient.last_tools == ""
+    assert agent.handler is None
+
+
+def test_reset_does_not_touch_pid_log_when_logging_is_disabled(tmp_path, monkeypatch):
+    pid_log = tmp_path / f"model_responses_{os.getpid()}.txt"
+    pid_log.write_text(LOG_CONTENT, encoding="utf-8")
+    monkeypatch.setattr(continue_cmd, "_LOG_DIR", str(tmp_path))
+
+    continue_cmd.reset_conversation(_agent(False))
+
+    assert pid_log.read_text(encoding="utf-8") == LOG_CONTENT
+    assert not list(tmp_path.glob("model_responses_snapshot_*.txt"))
+
+
+def test_snapshot_retains_the_legacy_pid_fallback(tmp_path, monkeypatch):
+    pid_log = tmp_path / "model_responses_123456.txt"
+    pid_log.write_text(LOG_CONTENT, encoding="utf-8")
+    monkeypatch.setattr(continue_cmd, "_LOG_DIR", str(tmp_path))
+
+    snapshot = continue_cmd._snapshot_current_log(123456)
+
+    assert pid_log.read_text(encoding="utf-8") == ""
+    assert Path(snapshot).read_text(encoding="utf-8") == LOG_CONTENT