From a9fb3d6591a3601d4228747ed17d899fb348155b Mon Sep 17 00:00:00 2001
From: Mark Barney <82deutschmark@gmail.com>
Date: Sat, 7 Mar 2026 01:26:08 -0500
Subject: [PATCH] fix: suppress thinking-mode output in ReviewPlanTask system
 prompt

Qwen 3.5-35B-A3B enters chain-of-thought reasoning mode on complex
multi-turn prompts, outputting a large 'Thinking Process:' block
before the JSON. This exhausts the num_output token budget, leaving
no tokens for the actual JSON response. JSON extraction then fails
with 'Could not extract json string from output'.

Fix: explicit instruction to output ONLY the JSON object without
any reasoning preamble.

Confirmed failure: ReviewPlanTask failed 2026-03-07 01:05 EST
on Mac Mini M4 Pro 64GB, Qwen 3.5-35B-A3B, num_output=8192.
---
 worker_plan/worker_plan_internal/plan/review_plan.py | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/worker_plan/worker_plan_internal/plan/review_plan.py b/worker_plan/worker_plan_internal/plan/review_plan.py
index 61810449..4034b604 100644
--- a/worker_plan/worker_plan_internal/plan/review_plan.py
+++ b/worker_plan/worker_plan_internal/plan/review_plan.py
@@ -49,6 +49,8 @@ class DocumentDetails(BaseModel):
 Do not include any extra bullet points, header lines, or any additional text outside of this JSON structure.
 
 Do not duplicate issues already identified in previous questions of this review.
+
+Output ONLY the JSON object. Do not include any thinking process, reasoning steps, or analysis before the JSON. Begin your response immediately with the opening brace {.
 """
 
 @dataclass