setup-loop_skill

Overview

This skill sets up a Ralph execution loop in Beads by converting a DevAgent plan into an epic and its child tasks with the correct fields, dependencies, and routing labels. It guarantees traceability by embedding the plan path in epic descriptions and task notes and enforces routing rules so Ralph can assign work to agents reliably. Use it to prepare Beads for safe, resumable autonomous execution.

How this skill works

It reads a drafted DevAgent plan and generates one epic plus direct child tasks, applying exact Beads field mappings: title, description (including plan path, final deliverable, and quality gates), task descriptions, acceptance criteria, and notes. It enforces routing by adding exactly one routing label per direct epic child, creates issues first and links dependencies in a second pass, and appends a final "Generate Epic Revise Report" project-manager task that depends on all top-level tasks. The process is idempotent and designed to be retried from any interruption.

When to use it

• After a plan is drafted and approved, before invoking Ralph execution
• When Beads metadata, routing labels, or agent assignments must be correct
• When dependencies and parent linkage need to be established reliably
• When you must embed plan references, quality gates, and acceptance criteria into Beads fields

Best practices

• Create the epic and each direct child task individually, then add dependency edges in a separate step
• Ensure each direct epic child has exactly one routing label from the Ralph config mapping
• Include the absolute plan path in the epic description and every task's notes for traceability
• Detect quality gates from package.json and list them in the epic—do not invent gates
• Prefer hierarchical numeric IDs but explicitly set the parent field after creation to ensure filters work

Example use cases

• Converting a finalized DevAgent plan into a Beads epic with labeled tasks so Ralph can route work to engineering, design, QA, or PM agents
• Preparing a multi-role UI feature loop: add design checkpoint tasks when UI-sensitive files or keywords are present
• Creating a stepwise, resumable setup for a complex epic so interruption-safe dependency linking can be retried
• Adding the mandatory post-work report task that depends on all top-level tasks to enforce quality gate review