[Bug]: Content wiping after using removeThinkTags step 2 (Reported by Nanobrowser)

Nanobrowser Version

0.1.13

Bug Description

When using the removeThinkTags step 2, legitimate content is being wiped out after a stray </think> /> tag is processed. This leads to data loss and corrupted output.

Expected Behavior: The operation should remove only the specified tags without corrupting the surrounding legitimate content.

Actual Behavior: Legitimate content is wiped out, leading to data loss.

Steps to Reproduce

1. Execute the process involving removeThinkTags step 2.
2. Introduce a stray closing think tag: </think> /> in the input.
3. Observe the resulting content after the operation

LLM Service Provider

Bionic (LM Studio)

Models Used

Planner: google/gemma-4-e2b Navigator: google/gemma-4-e2b Validator: -

Screenshots