Introduction to Ralphy

In the age of rapid dev cycles, developers are constantly looking for ways to cut down on repetitive coding chores. Ralphy offers a unique solution: an autonomous command‑line interface (CLI) that automatically loops through a series of AI engines—Claude Code, Codex, OpenCode, Cursor, Qwen, Factory Droid, and GitHub Copilot—to turn a product requirement document (PRD) into actual code.

Why Ralphy Stands Out

• Multi‑Engine Support – Switch between AI models on the fly or run them in parallel for faster feedback.
• Task Automation – A single command can spin up a fully functional feature, or you can feed an entire PRD with tasks already listed.
• Smart Conflict Resolution – When agents run in parallel, Ralphy automatically merges branches and resolves most conflicts.
• Browser Automation – Use agent‑browser to test UI flows or verify deployments right from the terminal.
• Customizability – Define project rules, model overrides, and even webhook notifications for Discord, Slack, or custom services.
• Lightweight Sandboxing – For monorepos with huge node_modules, Ralphy offers sandbox mode to avoid duplicate dependency copies.

Getting Started

Installation (npm)

npm install -g ralphy-cli

Installation (Git Clone)

git clone https://github.com/michaelshimeles/ralphy.git
cd ralphy
chmod +x ralphy.sh
./ralphy.sh "add login button"

Both installation methods expose the same command set; the shell script is handy for environments that don’t have Node.js.

Usage Scenarios

Single Task

ralphy "add dark mode"
ralphy "fix the auth bug"

Full PRD

Place a Markdown file called PRD.md in your repo root:

# Product Requirement Document

## Tasks
- [ ] create auth
- [ ] add dashboard
- [ ] setup CI/CD

Then run:

ralphy --prd PRD.md

Configure the Project

ralphy --init   # auto‑detects settings like language, framework
ralphy --add-rule "use TypeScript strict mode"   # add custom rule to config
ralphy --config   # view current configuration

Parallel Engines

Speed up work by running multiple agents at once:

ralphy --parallel          # 3 agents by default
ralphy --parallel --max-parallel 5   # increase pool

Branch‑Per‑Task Workflow

ralphy --branch-per-task --create-pr --draft-pr

This creates a feature branch for each task (e.g., ralphy/create-auth), merges changes automatically, and can even open draft PRs.

Advanced Features

• Model Override – ralphy --model sonnet "add feature" or ralphy --sonnet "add feature".
• Engine‑Specific Arguments – Pass CLI flags after -- to the underlying engine.
• Webhook Notifications – Add discord_webhook, slack_webhook, or custom_webhook URLs in .ralphy/config.yaml for real‑time updates.
• Browser Automation – Enable with --browser; Ralphy will invoke agent-browser to perform clicks, type into fields, and take screenshots.
• Sandbox Mode – Use --sandbox for faster clones that symlink large dependency trees.

Getting Help and Contributing

The project is open to contributions. Join the community on Discord, submit pull requests, or open issues on GitHub. The repository’s README contains detailed docs, example PRDs, and changelog entries.

Conclusion

Ralphy bridges the gap between product specs and code, automating most of the grunt work that slows down development. By leveraging multiple LLMs, parallel execution, and auto‑merged PRs, it turns a simple PRD into actual, runnable features—all from the comfort of your terminal. Whether you’re a solo developer wanting to speed things up or a team looking to standardize AI‑driven coding, Ralphy is a tool worth adding to your dev stack.