How does the complexity routing work?

The skill analyzes your task and assigns a score between 0 and 1.0, then selects an appropriate strategy such as Direct Execution for simple tasks or Autonomous Evolution for complex ones.

Can I set my own quality requirements?

Yes, you can use the --threshold parameter to define exactly how high the quality score must be (e.g., 0.95) before the iteration loop completes.

Is this skill limited to specific programming languages?

No, it is language-agnostic and can be applied to any task including coding, architectural design, documentation, or complex logic verification.

What is the benefit of the .prompts directory?

It automatically saves every iteration's prompt, response, and quality score, allowing you to audit the evolution of the solution and revert to previous versions if needed.

Meta-Prompt Iteration Framework

Name: Meta-Prompt Iteration Framework
Author: HermeticOrmus

byHermeticOrmus

•

Productivity & Workflow

Recursively refines AI-generated code and content through automated quality assessment and complexity-aware prompting.

The Meta-Prompt Iterate skill implements a sophisticated meta-cognitive workflow for Claude Code, designed to move beyond simple one-shot prompts. It automatically analyzes the complexity of a task to select the best prompting strategy, generates solutions, extracts context from previous attempts, and iteratively improves the output until a specific quality threshold is met. By archiving every iteration in a structured directory, it provides a transparent audit trail of the AI's reasoning process, making it indispensable for production-critical code, complex architectural designs, and high-precision technical tasks.

Key Features

01Pattern and constraint extraction from previous output iterations

02Automated complexity routing for Simple, Medium, and Complex tasks

03Structured project archiving in specialized .prompts directories

044 GitHub stars

05Customizable quality thresholds and domain-specific personas

06Recursive quality assessment with numeric scoring and reasoning

Use Cases

01Architecting complex distributed systems through iterative hypothesis testing

02Generating production-grade code with comprehensive edge-case handling

03Systematic refinement of technical documentation and implementation patterns

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add hermeticormus/claude-code-game-development meta-prompt-iterate

For use in Claude.ai and ChatGPT

Download Skill