[BUG] --dangerously-skip-permissions should also skip workspace trust dialog (or add --trust flag)

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?

When launching Claude Code with --dangerously-skip-permissions and agent teams (--teammate-mode tmux), the workspace trust dialog ("Do you trust the files in this folder?") still appears — once for the main session AND once for each spawned teammate agent.

What Should Happen?

Expected behavior:

If I've already opted into --dangerously-skip-permissions (which explicitly says "bypass all permission checks"), the workspace trust dialog should also be bypassed — or at minimum, trust should propagate from the parent session to spawned teammates.

Error Messages/Logs

Steps to Reproduce

Reproduction:

  CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS=1 claude \
    --teammate-mode tmux \
    --model "claude-opus-4-6[1m]" \
    --effort max \
    --dangerously-skip-permissions

Then use the Agent Company pattern (or any multi-agent workflow). Each teammate agent requires manual approval of the trust dialog before it can begin work.

Claude Model

Opus

Is this a regression?

No, this never worked

Last Working Version

_No response_

Claude Code Version

2.1.84

Platform

Anthropic API

Operating System

macOS

Terminal/Shell

iTerm2

Additional Information

Suggested fix (any of these):

1. Have --dangerously-skip-permissions also skip workspace trust (it already implies full trust)
2. Add a --trust CLI flag to skip just the workspace trust dialog
3. Auto-propagate workspace trust from parent to teammate/subagent processes
4. Add a trustedDirectories array in settings.json for pre-trusting paths

Impact:

Multi-agent workflows (agent teams, Agent Company) are effectively blocked without manual intervention at each agent's trust prompt, defeating the purpose of autonomous agent teams.