Hello,
I am writing to express my extreme dissatisfaction with the performance of the Antigravity client and the provided AI models.
Issue Summary:
The models consistently demonstrate a complete inability to self-configure and orchestrate their built-in tools (skills). During operation, the AI chaotically triggers internal functions and then crashes or terminates the session, citing context, token, or API limits. This is completely absurd.
Key Complaints:
- Total lack of autonomy: Instead of solving the assigned tasks automatically, the model breaks its own execution logic and forces the user to deal with system-level errors.
- Useless tools (skills): The integrated skills function incorrectly, causing infinite loops and exhausting limits before delivering any useful output.
- Wasted resources: Due to these internal system bugs, available quotas and limits are completely burned through while the task remains unaccomplished.
Demand:
Please forward this complaint to the development team to optimize how tool calling and skills are handled in the Antigravity client. As it stands, this “automation” is completely broken and useless. The best immediate solution is to provide an option to completely disable built-in skills/tools in the settings, allowing the model to run in a pure text-only mode without these critical crashes.
I look forward to your response and an update on the measures taken.