fix: follow-ups for salvaged PR #72425

- codex app-server sibling path: surface a WARNING (was silent debug) when
  the projected-message flush fails — same bug class as the main fix, but
  codex output has already streamed so fail-closed and agent_persisted=False
  are both wrong here (#860/#42039 duplicate-write hazard); loud durability
  gap logging instead.
- map session_persistence_failed in _format_turn_completion_explanation so
  the user sees an actionable reason instead of 'The request failed:
  unknown error' + explainer test.
- contributors/emails mapping for elco@thedaoist.gg (attribution CI).

agent/codex_runtime.py

@@ -778,12 +778,27 @@ def run_codex_app_server_turn(
         # the already-flushed user turn). See gateway/run.py agent_persisted.
         if getattr(agent, "_session_db", None) is not None:
             try:
-                agent._flush_messages_to_session_db(messages)
+                _codex_flush_ok = agent._flush_messages_to_session_db(messages)
             except Exception:
-                logger.debug(
+                _codex_flush_ok = False
+                logger.warning(
                     "codex app-server projected-message flush failed",
                     exc_info=True,
                 )
+            if _codex_flush_ok is False:
+                # Unlike the chat-completions loop (which fails closed BEFORE
+                # projection — see conversation_loop session_persistence_failed),
+                # codex output has already streamed to the user by the time this
+                # flush runs, so there is nothing left to withhold. We cannot
+                # flip agent_persisted=False either: the gateway fallback write
+                # would re-INSERT the already-flushed user turn (#860/#42039).
+                # Surface the durability gap loudly instead of a silent debug.
+                logger.warning(
+                    "codex app-server turn was delivered but could NOT be "
+                    "persisted to the session DB (session=%s) — this turn "
+                    "will be missing after restart/resume",
+                    getattr(agent, "session_id", None),
+                )
 
 
     # Counter ticks for the agent-improvement loop.

contributors/emails/elco@thedaoist.gg

@@ -0,0 +1 @@
+elcocoel

run_agent.py

@@ -3445,6 +3445,14 @@ class AIAgent:
                 "the model produced no follow-up text. Send `continue` to "
                 "let it summarize."
             )
+        if reason == "session_persistence_failed":
+            return (
+                prefix
+                + "the turn was stopped because session storage could not be "
+                "written (the transcript would have been lost on restart). "
+                "Check disk space / permissions for the state DB, then send "
+                "your message again."
+            )
         # Unknown/diagnostic-only reasons (e.g. "unknown", guardrail_halt
         # which already surfaces its own message) — don't second-guess.
         return ""

tests/run_agent/test_turn_completion_explainer.py

@@ -106,6 +106,16 @@ def test_explanation_for_all_retries_exhausted():
     assert "retries" in out.lower()
 
 
+def test_explanation_for_session_persistence_failed():
+    """Fail-closed persistence exits (#72425) must explain themselves."""
+    out = AIAgent._format_turn_completion_explanation(
+        "session_persistence_failed"
+    )
+    assert out  # non-empty
+    assert "session storage" in out.lower()
+    assert "disk space" in out.lower()
+
+
 # --------------------------------------------------------------------------
 # 2. Enable/disable seam
 # --------------------------------------------------------------------------