VS Code extension: "Trust mode" or equivalent of --dangerously-skip-permissions

Resolved 💬 3 comments Opened Mar 19, 2026 by ojhurst Closed Apr 16, 2026

When I use Claude Code in the terminal with --dangerously-skip-permissions, it's a completely different experience. Claude flows — it reads files, edits code, runs commands, and chains actions together without stopping me every 10 seconds to approve something I was obviously going to approve.

In the VS Code extension, that flow doesn't exist. Every shell command, every file write, every tool call hits a permission prompt. For power users who already trust Claude with their codebase (and have chosen to run it with broad access), this friction kills the magic.

What I'd like

A way to run the VS Code extension in a mode where tool calls execute without individual approval — the same trust level as the terminal's --dangerously-skip-permissions flag. This could be:

  • A setting in the extension config (e.g., claude-code.skipPermissions: true)
  • A toggle in the UI (like a "Trust mode" switch)
  • Respecting the same --dangerously-skip-permissions flag from .claude/settings.json

Why this matters

The whole value of Claude Code is that it's an agent that does things. When every action requires a click to approve, it becomes a suggestion engine with extra steps. The terminal version already solves this — the extension should offer parity.

I understand the safety rationale. A warning on first enable is fine. But let me opt in once and get out of Claude's way.

View original on GitHub ↗

This issue has 3 comments on GitHub. Read the full discussion on GitHub ↗