Built-in Plan agent ignores parent settings.json permissions and repeatedly prompts for pre-approved tools

[PaulRBerg]

Description

The built-in Plan agent repeatedly asks for permissions for tools that are already approved in my root ~/.claude/settings.json file. This makes the Plan agent workflow frustrating and interrupts the planning process.

Environment

• Claude Code Version: 2.0.28+ (Plan agent introduced in this version)
• Platform: macOS (Darwin 25.0.0)
• Configuration: Extensive permissions.allow list in ~/.claude/settings.json

Current Behavior

When Claude Code invokes the built-in Plan agent (subagent), it repeatedly requests permissions for Bash tools that are already approved in my root settings.json file.

I've noticed that it happens most often with piped (|) commands.

Expected Behavior

Built-in subagents like Plan should inherit or respect the permissions already configured in settings.json, similar to how the main Claude conversation respects these settings.

Root Cause Analysis

Based on documentation review:

1. Subagents have their own separate tool permissions
2. The tools field inheritance (when omitted) only applies to custom subagents
3. Built-in subagents like Plan have predefined tool configurations that don't inherit from parent settings
4. Users cannot customize built-in subagent permissions

Related Issues

This issue is related to but distinct from:

• [BUG] Task subagents fail to inherit permissions in MCP server mode (affects both WSL and native Windows) #5465 - Task subagents fail to inherit permissions (MCP-specific)
• [BUG] Need better way to restrict subagent tool use #4801 - Need better way to restrict subagent tool use
• Ambiguity in Subagent Behavior When Main Agent is in Plan Mode #4750 - Ambiguity in Subagent Behavior in Plan Mode
• [FEATURE] User-Level Agent Rules and Rule Propagation #8395 - User-Level Agent Rules and Rule Propagation

However, this specifically affects built-in agents like Plan, not just custom subagents or MCP scenarios.

Proposed Solution

1. Option A: Built-in subagents should inherit permissions from parent settings.json by default
2. Option B: Provide a way to configure built-in subagent permissions (e.g., .claude/agents/plan.override.md)
3. Option C: Add a setting like "builtinAgentsInheritPermissions": true

Workarounds Attempted

None available. Cannot modify built-in agent configurations.

Impact

• Interrupts workflow with repeated permission prompts
• Makes Plan mode less useful despite being a valuable feature
• Undermines the purpose of pre-configuring extensive permission lists

[PaulRBerg]

For example:

Even if I have find and head in permissions.allow.

[brendanstennett]

+1 Happening consistently for me as well. Have looked for syntax errors/typos in settings.local.json (which makes the whole file get silently ignored, but that's another problem) and --dangerously-skip-permissions, neither of which allow the plan agent to operate without checking for every permissions.

This is a critical bug IMO because the expected workaround for users is to opt in to --dangerously-skip-permissions which we know is, as the flag states, dangerous. This is potential liability for Anthropic.

[PaulRBerg]

@brendanstennett I agree this should be treated as a critical bug.

Until they fix it, I was able to fix this by adding this permission:

{
  "permissions": {
    "allow": [
      "Bash"
    ]
  }
}

[brendanstennett]

Thanks @PaulRBerg

Sadly, just a "Do anything you want with Bash" is just a very risky workaround. I've been using devcontainers to help alleviate the risks associated as well.

[PaulRBerg]

I know it's risky. But the alternative is to give 200+ approvals every day for commands that I have already approved. That would be a massive slowdown in productivity.

I urge the Anthropic team to fix this bug ASAP 🙏