From 156b50358b14d170985bd66b02b57eeea030442d Mon Sep 17 00:00:00 2001
From: Teknium <127238744+teknium1@users.noreply.github.com>
Date: Wed, 25 Mar 2026 18:57:18 -0700
Subject: [PATCH] fix(reasoning): skip duplicate callback for <think>-extracted
 reasoning during streaming (#3116)
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Local models (Ollama, LM Studio) embed reasoning in <think> tags in
delta.content. During streaming, _stream_delta() already displays these
blocks. Then _build_assistant_message() extracts them again and fires
reasoning_callback, causing duplicate display.

Track whether reasoning came from structured fields (reasoning_content)
vs <think> tag extraction. Only fire the callback for <think>-extracted
reasoning when stream_delta_callback is NOT active. Structured reasoning
always fires regardless.

Salvaged from PR #2076 by dusterbloom (Fix A only — Fix B was already
covered by PR #3013's _current_reasoning_callback centralization).
Closes #2069.
---
 run_agent.py | 14 ++++++++++----
 1 file changed, 10 insertions(+), 4 deletions(-)

diff --git a/run_agent.py b/run_agent.py
index 8f62a3b2..28519905 100644
--- a/run_agent.py
+++ b/run_agent.py
@@ -4494,6 +4494,7 @@ class AIAgent:
         so both the tool-call path and the final-response path share one builder.
         """
         reasoning_text = self._extract_reasoning(assistant_message)
+        _from_structured = bool(reasoning_text)
 
         # Fallback: extract inline <think> blocks from content when no structured
         # reasoning fields are present (some models/providers embed thinking
@@ -4509,10 +4510,15 @@ class AIAgent:
             logging.debug(f"Captured reasoning ({len(reasoning_text)} chars): {reasoning_text}")
 
         if reasoning_text and self.reasoning_callback:
-            try:
-                self.reasoning_callback(reasoning_text)
-            except Exception:
-                pass
+            # Skip callback for <think>-extracted reasoning when streaming is active.
+            # _stream_delta() already displayed <think> blocks during streaming;
+            # firing the callback again would cause duplicate display.
+            # Structured reasoning (from reasoning_content field) always fires.
+            if _from_structured or not self.stream_delta_callback:
+                try:
+                    self.reasoning_callback(reasoning_text)
+                except Exception:
+                    pass
 
         msg = {
             "role": "assistant",