Add Ralph autonomous agent loop engine with Copilot backend

[Copilot]

Implements the Ralph pattern from https://github.com/snarktank/ralph - an autonomous agent loop that repeatedly executes Copilot until all PRD user stories are complete. Each iteration spawns a fresh agent instance with clean context; memory persists via git history, progress.txt, and prd.json.

Implementation

New engine type: ralph

• Wraps Copilot CLI with autonomous loop orchestration
• Delegates installation, MCP config, and log parsing to Copilot engine
• Manages PRD-driven task execution with JSON user stories

Loop mechanics (actions/setup/sh/ralph-loop.sh):

• Iterates until all stories pass or max iterations reached
• Each iteration: select incomplete story → execute Copilot → update status → commit
• State tracked in /tmp/gh-aw/ralph/{prd.json,progress.txt}

Schema changes:

• Added ralph to engine enum in main_workflow_schema.json
• Registered in global engine registry alongside claude/codex/copilot/custom

Usage

---
engine: ralph
on: workflow_dispatch
permissions:
  contents: read
---

{
  "userStories": [
    {
      "id": "story-1",
      "title": "Create hello.py",
      "passes": false
    }
  ]
}

The loop automatically picks incomplete stories, implements them via Copilot, validates, and marks complete.

Original prompt

Add ralf loop + copilot as a custom engine

https://github.com/snarktank/ralph

---

💡 You can make Copilot smarter by setting up custom instructions, customizing its development environment and configuring Model Context Protocol (MCP) servers. Learn more Copilot coding agent tips in the docs.