[Bug] The permissions are too strict, resulting in failed tool calls without any approvals in the "Auto Review" mode.

Open 💬 0 comments Opened May 30, 2026 by roflsunriz

What version of the IDE extension are you using?

26.527.31454

What subscription do you have?

ChatGPT Plus

Which IDE are you using?

VS Code 1.122.0

What platform is your computer?

Microsoft Windows NT 10.0.26200.0 x64

What issue are you seeing?

Unless an agent is explicitly instructed to "promote and execute", all tool queries in the auto review mode will fail, causing the agent to become stagnant.

What steps can reproduce the bug?

Select the "Auto Review" mode and give instructions to the agent.

What is the expected behavior?

Basic tools such as "edit", "read", "patch", "webfetch", and "websearch" should be allowed from the start (just like in previous versions).

Additional information

As a drawback, using tools requires time for review and approval, which in turn increases the execution time.

View original on GitHub ↗