Infinite Loop in API Request generation via Gemini AI Studio "Build" feature

[scorpioyuumei]

Hello,

I am reporting a potential bug regarding the code generated by the “Build” feature in Gemini AI Studio.

Summary of the issue: When using the “Build” feature to generate a functional application, the resulting code logic contains a structural flaw that leads to an unintended infinite loop of API calls.

Steps to reproduce:

1. Open Gemini AI Studio and select the “Build” feature.

2. Provide a prompt to generate an application that requires sequential processing.

3. Upon executing the generated code, the API request logic fails to properly check exit conditions, resulting in continuous, rapid-fire API calls until the execution environment is manually terminated.

Expected behavior: The generated code should include proper exit conditions or rate limiting to prevent uncontrolled API request loops.

Actual behavior: The code generates an infinite loop that continues until an external intervention occurs, which poses a significant risk to execution stability.

I would appreciate it if the technical team could investigate the logic generation process of the “Build” feature and implement a fix for these recurring loop conditions.

Thank you for your support.

[Mustan_lokhand]

Do you experience this loop when using the app in AI Studio or when usign the code generated by the app ?

[scorpioyuumei]

Yes, that is correct. This issue occurred when I used the code generated by the “Build” feature in Gemini AI Studio. I did not intentionally include any loops; the infinite loop is a result of the logic within the generated code itself.

[Mustan_lokhand]

Can you share the prompt you used via the DM with me
I am working on reproducing this but am unable to do it with my prompts

[Chris_Malatesta]

Issue with Google AI Studio: Google AI Studio

[scorpioyuumei]

Mustan,

I apologize, but I do not have a record of the exact prompt I used at that time, so I cannot share it directly. However, I can provide a detailed description of the type of application I was trying to create and the API behavior I experienced. If there is any other information that would be helpful for your investigation, please let me know, and I will be happy to provide it.

[scorpioyuumei]

Hello, Mustan.

Thank you for your ongoing investigation. As I do not have the specific prompt history, I have compiled a summary of the application’s functionality and the technical issues I encountered, along with a video demonstrating the error.

[Summary Table]

Period	Issue	Cause
Early June	Generation state did not reset	Inconsistency in generation flag management (isGeneratingRef), causing the generation process to get stuck.
Mid-June	Mismatch in generation parameters (video duration)	Invalid video durations were selectable, causing API request errors and process termination.
Late June	Prompt structure errors leading to re-generation	The “7R” prompt template did not match the expected system structure, requiring template revisions.
Late June - Early July	Excessive retries on API timeouts	The timeout threshold (circuit breaker) for API response delays was incorrectly configured, causing intermittent process termination.

I hope this helps your team identify the root cause. Please let me know if you need any further technical details.