[FEATURE] Claude should proactively suggest updating settings.json permissions when a user asks about prompting for a tool repeatedly

Resolved 💬 2 comments Opened May 22, 2026 by gregwilliams-zimmer Closed May 26, 2026

Preflight Checklist

  • [x] I have searched existing requests and this feature hasn't been requested yet
  • [x] This is a single feature request (not multiple features)

Problem Statement

Title: Claude should proactively add repeated commands to permissions.allow

Body:

Summary

When a user repeatedly runs the same command and hits confirmation prompts, Claude should proactively offer to add the command to permissions.allow in ~/.claude/settings.json — and make the edit when the user agrees.

Current behavior

Claude never suggests the permissions.allow setting on its own. Users must explicitly ask how to avoid confirmation prompts, at which point Claude may describe the option but still does not apply the fix.

Expected behavior

When Claude observes that a user is repeatedly approving the same command, it should:

  1. Proactively offer to add it to permissions.allow
  2. Make the edit to ~/.claude/settings.json immediately when the user agrees — not just describe how to do it

Why this matters

A user reported asking multiple times across sessions for a way to run Gradle tests without confirmation prompts. The fix was never surfaced until they asked directly. The intent is obvious from context — a command run repeatedly in a development workflow should not require repeated approval.

Proposed Solution

Suggested fix

  • When Claude runs the same Bash command more than once in a session requiring confirmation, proactively suggest adding it to permissions.allow
  • When a user asks how to avoid confirmation prompts for a command, apply the settings edit directly rather than describing it

Alternative Solutions

_No response_

Priority

Medium - Would be very helpful

Feature Category

Interactive mode (TUI)

Use Case Example

Asking claude repeatedly not to prompt for gradle test runs yet claude still continues to prompt.

Additional Context

_No response_

View original on GitHub ↗

This issue has 2 comments on GitHub. Read the full discussion on GitHub ↗