Overview
A clean, independently authored set of prompt templates designed to slot directly into any Claude Code-based coding agent. Especially strong on safety architecture — reversibility tiers, destructive action gates, and risk assessment frameworks.
Features
- System Prompt Architecture: Layered identity + safety + behavior prompt composition.
- Safety & Risk Assessment: Reversibility tiers for every tool action category.
- Tool Routing: Custom routing logic for directing tasks to the right subagent.
- Multi-Agent Coordination: Handoff protocols, context isolation, and delegation templates.
Use Cases
- Adding professional-grade safety rails to your Claude Code pipeline.
- Building a custom orchestrator with proper reversibility enforcement.
- Drop-in templates for any team adopting Claude Code as a shared developer tool.
Technical Advantages
- Written independently of Anthropic — unbiased analysis of what actually works.
- Templates are format-agnostic: works with CLAUDE.md, skills, and slash commands.