In the build section of Google ai studio when I use a model like Gemini 3.5 or 3.6 The Toolcall doesn’t get through I tell it Lets for example I tell it to view a file and do some edits
It does a series of tool calls When it finishes it tells you it’s done, no This time Like I tell it view my file It responds to me "I will view file_name and inspect what’s going on " the message end no toolcalls
Hi @Mostafa_s_Academy1 ,
Make sure your chat history length isn’t getting excessive. Do your best to start a new chat when you begin working on a new part of your codebase or project. I notice the agent can become confused when chats get lengthy. First they read everything up until your new prompt, then the agent has to decide if everything it read up until that point is related to the actual request you’re making. This can confuse the agent and cause it to not respond correctly.
Also, when you’re making edits to specific files, make sure you’re typing the exact file name of the file you would like to edit. Most importantly, if you’re not sure which file the edit you would like to accomplish is associated with, don’t guess. Instead - describe where in the project the edit should be performed using visual cues (On the landing page, at the bottom of the page, in the container with these buttons, etc). If you accidentally guess the wrong file and tell the agent to edit what isn’t there, it might refuse to make the edit. It always assumes you’re correct, and might not argue with you about the change.
I hope this helps!
-TechRay
Thank you very much for yt advice