automatic mode is broken for ´run_terminal_command´ tool

Before submitting your bug report

• I've tried finding an answer on the Continue docs site
• I'm not able to find an open issue that reports the same bug
• I've seen the troubleshooting guide on the Continue Docs

Relevant environment info

- OS: Windows 11
- Continue version: 2.0.0
- IDE version: 1.129.1
- Model: Claude Sonnet 4.5 / Claude Opus 4.8, GPT-5.5, ...

Description

Built-In tools support setting the mode to Automatic. Setting run_terminal_command to Automatic has no effect. All terminal commands still require manual confirmation. To my knowledge, other tools work in Automatic mode.

This functionality has been broken since at least March 2026 when I first tried Continue.dev - and discarded it for this reason. Setting run_terminal_command to automatic has no effect!

Verified again today with Continue.dev extension version 2.0.0, VSCode 1.129.1 on Win 11.

This issue was already reported in #12500 and #5353 which shows that is a problem EVERY user has. This is a critical issue for many users obviously and needs to be fixed asap.

To reproduce

• Set run_terminal_command to Automatic
• Run a session in Agent mode that requires terminal commands to be executed
• You will be prompted to approve every single terminal command, no matter how trivial they are

Log output

Start by tracing the built-in run_terminal_command tool from its Automatic-mode setting through the terminal confirmation path. Reproduce the Windows 11 case in an Agent-mode session and compare it with other tools that honor Automatic mode. Done means terminal commands configured as Automatic no longer prompt for manual approval, with the relevant behavior covered by a test if the project provides one.