superloopy-loop

What problem does it solve?

This Skill automates the process of creating strict evidence loops for Codex and Claude Code tasks, ensuring durable progress, criteria, and artifact-backed completion.

Core Features & Use Cases

• Strict Evidence Loop Creation: Automatically create and manage evidence loops for tasks requiring durable criteria and artifact-backed completion.
• Loop Engineering: Supports loop engineering with commands for guidance, proof, preflight, and finishing loops.
• Crew Mode: Escalate to full fan-out, dispatching across independent lanes with role-specific agents for implementation, review, QA, and audit.
• Optional Subagent-Driven Mode: Utilize custom agents from .codex/agents/ for subagent-driven workflows.
• Evidence Audit: Independently re-checks recorded proof and ensures criteria are met with a deterministic spine.
• Quality Gate: Ensures all artifacts exist, are non-empty, and live under the active evidence root.
• Steering: Accepts structured prompt-level steering for adding goals, revising criteria, and reordering pending goals.
• Doctor: Checks the package, hooks, skills, CLI, and runtime environment for issues.

Quick Start

Run 'loopy <task>' to initiate a strict evidence loop for your Codex or Claude Code task.