[FEATURE] Allow specifying additional paths for claude code artifacts (commands, agents, and hooks)

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

I have started assembling a collection of re-usable prompts, commands, and agents that I use for my own projects, and share with others. In order to implement these, I have to copy/paste them either into ~/.claude so they're available globally, or into each individual project's .claude directory. This can be a pain to manage.

Other solutions like symlinking to a checkout of the project make it hard to mix in the shared project's definitions with my own local custom definitions.

Proposed Solution

Allow additional paths for .claude files to be loaded from in settings.json and settings.local.json. When specified, these additional paths will be loaded in the order specified. Optionally, allow specifying a namespace for anything in the specified path to help resolve any naming conflicts.

Alternative Solutions

_No response_

Priority

High - Significant impact on productivity

Feature Category

Configuration and settings

Use Case Example

.claude/settings.local.json

{
  "paths": [ '~/agentic/007' ]
  "permissions": {
    "allow": [
      "Bash(mvn clean:*)",
      "Bash(mvn test:*)",
    ],
    "deny": []
  },
  "enabledMcpjsonServers": [
    "github",
    "playwright",
    "mysql"
  ]
}

Now with this configuration, ~/agentic/007 will be traversed to look for agents, hooks, and commands.

Additional Context

_No response_