Objective
Research the Ralph Loop pattern and document how it can be implemented using gh-aw agentic workflows.
Context
Ralph is an autonomous AI agent loop that runs AI coding tools repeatedly until all PRD items are complete. Each iteration is a fresh instance with clean context. Memory persists via git history, progress.txt, and prd.json.
Reference: https://github.com/snarktank/ralph
Approach
-
Review the Ralph Loop pattern and its key concepts:
- Fresh context per iteration
- Memory via git history + progress.txt + prd.json
- Small, atomic tasks
- Feedback loops (typecheck, tests)
- AGENTS.md updates for learnings
-
Identify mapping between Ralph concepts and gh-aw features:
- How to structure PRD as workflow frontmatter
- How to track task completion state
- How to persist learnings between iterations
- How to implement quality checks
-
Create a design document in docs/examples/ralph-loop.md covering:
- Overview of Ralph Loop pattern
- Mapping to gh-aw agentic workflows
- Proposed workflow structure
- Key differences from bash-based Ralph
Files to Create
docs/examples/ralph-loop.md - Design document
Acceptance Criteria
AI generated by Plan Command for #11132
Objective
Research the Ralph Loop pattern and document how it can be implemented using gh-aw agentic workflows.
Context
Ralph is an autonomous AI agent loop that runs AI coding tools repeatedly until all PRD items are complete. Each iteration is a fresh instance with clean context. Memory persists via git history,
progress.txt, andprd.json.Reference: https://github.com/snarktank/ralph
Approach
Review the Ralph Loop pattern and its key concepts:
Identify mapping between Ralph concepts and gh-aw features:
Create a design document in
docs/examples/ralph-loop.mdcovering:Files to Create
docs/examples/ralph-loop.md- Design documentAcceptance Criteria
Related to Add Ralph Loop examples for agentic workflows #11132