From 0a2d4a6eea796e9ac9a589939fa9ab452a2937ec Mon Sep 17 00:00:00 2001
From: kshitijk4poor <82637225+kshitijk4poor@users.noreply.github.com>
Date: Thu, 2 Jul 2026 16:59:42 +0530
Subject: [PATCH] docs(codex): clarify stale-floor docstring reflects the 10k
 gate

The helper docstring described the typical ~15-25k gateway payload but
read as if that were the trigger range; the floor actually engages above
10k tokens. Clarify the prose to match the gate.
---
 agent/chat_completion_helpers.py | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/agent/chat_completion_helpers.py b/agent/chat_completion_helpers.py
index 4e52b914a..76ab24d48 100644
--- a/agent/chat_completion_helpers.py
+++ b/agent/chat_completion_helpers.py
@@ -134,7 +134,9 @@ def openai_codex_stale_timeout_floor(est_tokens: int) -> float:
     Gateway/Telegram sessions routinely ship ~15–25k tokens of tools +
     instructions before the first user message. Subscription-backed Codex can
     legitimately spend several minutes in backend admission/prefill at that
-    size; the generic 90s non-stream stale default aborts healthy calls.
+    size; the generic 90s non-stream stale default aborts healthy calls. The
+    floor engages above 10k estimated tokens so those gateway-scale payloads
+    are covered; smaller requests keep the generic default.
     """
     if est_tokens > 100_000:
         return 1200.0