What is the purpose of the SUMMARY.md file?

Every prompt execution generates a SUMMARY.md file designed for human scanning. It includes a one-liner description, key findings, created files, and any blockers or required human decisions.

What is a Claude-to-Claude pipeline?

It is a multi-stage workflow where the output of one Claude prompt is specifically structured (often with XML and metadata) to be consumed and processed by a subsequent Claude prompt.

Can I use this to improve existing plans?

Yes, the 'Refine' purpose is specifically designed to take existing research or plan outputs and iterate on them based on new feedback or requirements.

How does the skill handle task dependencies?

The execution engine scans your .prompts/ directory for existing research or plan files and builds a dependency graph to determine if prompts should run sequentially or in parallel.

Where are the prompts stored?

All prompts and their resulting artifacts are stored in a standardized .prompts/ directory, organized by topic and purpose, ensuring full provenance of your AI workflows.

Claude Workflow & Meta-Prompt Architect

Name: Claude Workflow & Meta-Prompt Architect
Author: Eric-Ness

byEric-Ness

Productivity & Workflow

Automates the creation and execution of structured multi-stage Claude-to-Claude pipelines for complex research, planning, and implementation.

About

This skill optimizes Claude's performance in multi-step workflows by generating structured meta-prompts specifically designed for machine-to-machine communication. It automates the entire lifecycle of a prompt—from requirement gathering and chain detection to dependency-aware execution—while maintaining a clear, file-based history of research, plans, and implementations. By producing XML-structured outputs for agents and human-readable SUMMARY.md files for developers, it ensures seamless transitions between AI stages while providing full transparency and provenance for every decision made.

Key Features

Multi-stage workflow generation covering Research, Plan, Implement, and Refine stages.
Standardized .prompts/ directory structure for maintaining project history and versioning.
Structured XML and metadata output optimization for reliable Claude-to-Claude parsing.
Automatic SUMMARY.md generation for quick human review of AI-generated outcomes.
0 GitHub stars
Dependency-aware execution engine for running prompts sequentially or in parallel.

Use Cases

Automating multi-agent workflows where the output of a research agent must be parsed by a planning agent.
Developing complex software features that require distinct research and architectural planning phases.
Iterating on existing project strategies through automated refinement loops and version tracking.

About

Key Features

Multi-stage workflow generation covering Research, Plan, Implement, and Refine stages.
Standardized .prompts/ directory structure for maintaining project history and versioning.
Structured XML and metadata output optimization for reliable Claude-to-Claude parsing.
Automatic SUMMARY.md generation for quick human review of AI-generated outcomes.
0 GitHub stars
Dependency-aware execution engine for running prompts sequentially or in parallel.

Use Cases

Automating multi-agent workflows where the output of a research agent must be parsed by a planning agent.
Developing complex software features that require distinct research and architectural planning phases.
Iterating on existing project strategies through automated refinement loops and version tracking.