A Claude Code plugin that prepares human-developed codebases for agentic development.
When you fork a repository or inherit a codebase built by humans, this plugin guides Claude through a systematic process to make it agent-friendly:
- Analyzes the current repository structure
- Asks clarifying questions about your intended workflow
- Restructures folders and files for agent navigability
- Creates
CLAUDE.mdand/context/infrastructure - Updates documentation for agentic development
- Validates changes don't break existing functionality
If you have the danielrosehill marketplace configured:
/plugin install make-agent-friendly@danielrosehillClone and add as a local plugin:
git clone https://github.com/danielrosehill/Make-Agent-Friendly.git
/plugin marketplace add ./Make-Agent-Friendly
/plugin install make-agent-friendlyFor backwards compatibility, you can still copy just the command file:
mkdir -p ~/.claude/commands
curl -o ~/.claude/commands/make-agent-friendly.md \
https://raw.githubusercontent.com/danielrosehill/Make-Agent-Friendly/main/commands/make-agent-friendly.md- Navigate to the repository you want to prepare for agentic development
- Start Claude Code
- Run the command:
/make-agent-friendly
Claude will:
- Analyze your codebase
- Ask you a series of intake questions
- Work through a structured checklist to prepare the repository
- Create necessary infrastructure (
CLAUDE.md,/context/, etc.) - Validate changes before committing
After running the command, your repository will typically include:
your-repo/
├── CLAUDE.md # Project context for AI agents
├── context/ # External docs, API refs, MCP notes
│ └── ...
├── planning/ # (Optional) Decision records, discussions
│ └── ...
├── app/ or src/ # Application code (if reorganized)
│ └── ...
└── README.md # Updated with agentic workflow notes
This plugin prepares repositories for human-guided agentic development:
- Human: Controls planning, provides instructions, supervises
- Agent: Executes code changes, suggests improvements, navigates codebase
The goal is clear separation of concerns and maximum agent efficiency.
Make-Agent-Friendly/
├── .claude-plugin/
│ └── plugin.json # Plugin metadata
├── commands/
│ └── make-agent-friendly.md
├── foundational.md # Detailed guidance reference
└── README.md
Fork this repository and modify:
commands/make-agent-friendly.md- The main command promptfoundational.md- Reference documentation for the approach.claude-plugin/plugin.json- Plugin metadata
See foundational.md for the detailed guidance document this command is based on.
MIT