How to disable warning : This code contains potentially...

[rbolog]

Hi,
Is it possible to disable that warning ?

This code contains potentially dangerous commands. Please review and understand the code before running.

[uinstinct]

I think this would be a feature request

[Luke1th]

I am having the same issue and it is annoying.

[uinstinct]

@TyDunn what are your thoughts?

[TyDunn]

Maybe just remove this? I don't think other CLIs have this. Check with @sestinj on exact details

[Laransec]

#10229

I made another request here. You can remove the check in the files I referenced - then build the CLI and copy the new dist over the installed dist folder.

But then it auto updates and breaks your work.

[darfaz]

I understand the friction, but I'd actually advocate for keeping this warning (maybe with a per-workspace or per-language dismissal option rather than a global disable).

The warning exists because AI coding assistants can generate code that:

• Executes shell commands (os.system, subprocess, exec)
• Modifies files outside the project
• Downloads and runs external scripts
• Contains obfuscated payloads (especially if the model was influenced by prompt injection through context)

In agent mode especially, where Continue can autonomously run terminal commands, a malicious or hallucinated command could cause real damage.

A reasonable middle ground:

• Per-workspace trust settings — Mark a workspace as trusted to reduce warnings
• Allowlist specific patterns — If you frequently work with shell scripts or DevOps code, let users allowlist certain command patterns
• Severity levels — Distinguish between "this runs a shell command" (medium) and "this downloads from an external URL and executes it" (critical)

If you do disable it, at minimum keep logging what gets executed so you have an audit trail.