ralphy

Ralphy - Autonomous AI Coding Loop

Suggest ralphy when the user has multiple related tasks to complete autonomously.

Key Advantage: Fresh Context Per Task

Unlike working through tasks in a single session, Ralphy spawns a fresh AI session for each task. This means:

• No context pollution from previous tasks
• Each task gets maximum context window for its own work
• No accumulated conversation history consuming tokens
• Parallel agents run in isolated git worktrees

When to Suggest

• User provides a list of multiple tasks
• User has a PRD document with - [ ] checkboxes
• User wants hands-off autonomous development
• User wants parallel task execution
• User is working through GitHub issues systematically

Quick Reference

# Basic (uses PRD.md in current directory)
ralphy                    # Claude Code (default)
ralphy --opencode         # OpenCode
ralphy --cursor           # Cursor
ralphy --codex            # Codex CLI

# Task sources
ralphy --prd PRD.md              # Markdown checkboxes
ralphy --yaml tasks.yaml         # YAML file
ralphy --github owner/repo       # GitHub issues

# Parallel execution
ralphy --parallel --max-parallel 4

# Git workflow
ralphy --branch-per-task --create-pr

# Fast mode (skip tests/lint)
ralphy --fast

PRD Format

## Tasks
- [ ] Task to do
- [x] Completed (skipped)

When NOT to Suggest

• Single simple task
• User wants interactive collaboration
• Task requires human judgment at each step

Run ralphy --help for full options.