From 0e2873a77d221649f613cca266612406e63e5870 Mon Sep 17 00:00:00 2001
From: Teknium <127238744+teknium1@users.noreply.github.com>
Date: Thu, 21 May 2026 18:07:09 -0700
Subject: [PATCH] fix(computer_use): build summary once before aux-vision
 routing branch

The cherry-pick of #22891 (max_elements cap) reshuffled _capture_response
so summary was assigned inside both the multimodal and AX branches,
but #30126's aux-vision routing call (_route_capture_through_aux_vision)
fires BEFORE either branch and references the not-yet-bound name.

Compute summary once up-front, keep the AX-branch rebuild for the
truncation note.
---
 tools/computer_use/tool.py | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

diff --git a/tools/computer_use/tool.py b/tools/computer_use/tool.py
index ab9339f4b6c..82c6792eeb9 100644
--- a/tools/computer_use/tool.py
+++ b/tools/computer_use/tool.py
@@ -467,6 +467,12 @@ def _capture_response(cap: CaptureResult, max_elements: int = _DEFAULT_MAX_ELEME
     ]
     if element_index:
         summary_lines.extend(element_index)
+    # Multimodal and AX paths both reference `summary`; build it once up-front
+    # so the aux-vision routing branch (which fires before either path is
+    # selected) has a valid value to hand to _route_capture_through_aux_vision.
+    # The AX path appends the "truncated to N of M" note to summary_lines
+    # below and rebuilds; the multimodal path keeps this version untouched.
+    summary = "\n".join(summary_lines)
 
     if cap.png_b64 and cap.mode != "ax":
         # Decide whether to hand the screenshot to the auxiliary.vision
@@ -492,7 +498,6 @@ def _capture_response(cap: CaptureResult, max_elements: int = _DEFAULT_MAX_ELEME
         # The multimodal response carries the screenshot, not the AX
         # elements array, so a "response truncated to N of M elements"
         # note would be inaccurate — skip it on this branch.
-        summary = "\n".join(summary_lines)
         return {
             "_multimodal": True,
             "content": [