[BUG] --dangerously-skip-permissions still does not bypass workspace trust prompt (blocks agent team teammates)

Resolved 💬 4 comments Opened Mar 19, 2026 by unitedideas Closed Apr 16, 2026

Preflight Checklist

  • [x] I have searched existing issues and this hasn't been reported yet
  • [x] This is a single bug report (please file separate reports for different bugs)
  • [x] I am using the latest version of Claude Code

What's Wrong?

--dangerously-skip-permissions does not bypass the workspace trust prompt ("Quick safety check: Is this a project you created or one you trust?"). This was previously reported in #28506 but was auto-closed as
a duplicate of #18942 — which is a different bug (trust prompt appearing specifically in ~ due to settings.json being loaded as both user and project settings).

The core issue remains: a flag literally named "dangerously skip permissions" does not skip all permission prompts.

This is particularly problematic for agent teams. When the lead session spawns teammates, each teammate is a new Claude Code session that hits the trust prompt. Teammates inherit the lead's
--dangerously-skip-permissions flag, but since the flag doesn't actually skip the trust dialog, every spawned teammate gets stuck on an interactive prompt that can't be dismissed programmatically. This makes
agent teams unusable in non-git directories or any workspace that triggers the trust check.

What Should Happen?

--dangerously-skip-permissions should bypass the workspace trust prompt. The user has explicitly opted into "bypass all permission checks" — a secondary trust confirmation is redundant and breaks
automated workflows (agent teams, CI, scripted launches).

Error Messages/Logs

Steps to Reproduce

  1. Navigate to a directory without .git:

cd /tmp && mkdir test-workspace && cd test-workspace

  1. Launch Claude Code:

claude --dangerously-skip-permissions

  1. Observe the trust prompt still appears
  2. Now try spawning an agent team — each teammate hits the same prompt and blocks

Claude Model

Opus

Is this a regression?

No, this never worked

Last Working Version

_No response_

Claude Code Version

2.1.79 (Claude Code)

Platform

Anthropic API

Operating System

macOS

Terminal/Shell

iTerm2

Additional Information

Previous issues — none resolved this:

  • #28506 — exact same report, auto-closed as duplicate of #18942
  • #18942 — different bug (home dir trust prompt), closed
  • #21283 — trust decisions not persisting across sessions
  • #6797 — related trust prompt issues

View original on GitHub ↗

This issue has 4 comments on GitHub. Read the full discussion on GitHub ↗