plugin-development

Independent Plugin for Claude Code

An autonomous task-driven loop that works through your TASKLIST.md independently. Claude keeps working until all tasks are complete.

Installation

# Add the marketplace
/plugin marketplace add SuedePritch/Independent

# Install the plugin
/plugin install independent@jamespritchard-plugins

Usage

1. Create a TASKLIST.md

In your project, create .claude/TASKLIST.md:

- [ ] Refactor the authentication module
- [ ] Add unit tests for user service
- [ ] Update API documentation
- [ ] Fix the bug in checkout flow

2. Run the Independent Loop

/independent

Claude will:

• Work through each task autonomously
• Use appropriate tools and subagents for complex work
• Mark tasks complete as it finishes them
• Keep going until all tasks are done

3. Mark Tasks Complete

Tasks are tracked with checkboxes:

• - [ ] = pending
• - [x] = complete

Claude marks tasks complete automatically, or you can check them off manually.

How It Works

The plugin uses Claude Code hooks to enforce task completion:

Hook	Purpose
SessionStart	Clears session log for fresh starts
PreToolUse	Monitors task progress, enforces agent ownership
PostToolUse	Tracks agent invocations
Stop	Blocks exit if tasks remain incomplete
SubagentStart	Injects pending task context into subagents

When you try to exit with incomplete tasks, the Stop hook blocks it and shows remaining work.

Tips

• Break down complex tasks — Add subtasks to TASKLIST.md as you discover them
• Be specific — "Fix login bug" is better than "Fix bugs"
• Let it run — The loop is designed to work autonomously

Updating

Updates are automatic. When a new version is pushed, run:

/plugin install independent@jamespritchard-plugins

Plugin Structure

independent/
├── .claude-plugin/
│   └── plugin.json       # Plugin metadata
├── commands/
│   └── independent.md    # The /independent command
├── hooks/
│   ├── hooks.json        # Hook configuration
│   ├── session-start.sh  # Session initialization
│   ├── pretooluse-hook.sh
│   ├── post-tool-use.sh
│   ├── stop-hook.sh      # Exit gate
│   └── subagentstart-hook.sh
└── scripts/
    └── setup-independent.sh

License

MIT