Multi-Agent Skill Creator

This directory contains the Multi-Agent Skill Creator skill.

It is designed to help engineers create structured, coordinated multi-agent workflows (skills) for complex, repeatable tasks (coding or non-coding).

It operates under a Cold Logic mandate: providing honest, data-driven, and objective guidance to design optimal workflows, including advising when a multi-agent approach is unnecessary.

How to Use

1. Read SKILL.md to understand the principles and stages.
2. Invoke the skill by describing your target workflow to Gemini and asking for assistance in creating a multi-agent skill for it.
3. Participate in the interactive discovery and design phase.
4. Use the generated templates and validation scripts to bootstrap your new skill.

Verification & Objectives

To maintain the quality and consistency of this skill creator, it is equipped with its own validation infrastructure.

Validation Objectives

Any changes to this directory must satisfy the following criteria:

1. Markdown Integrity: All links in SKILL.md and README.md must be valid relative links. No broken links are allowed.
2. Template Correctness: Python template files in templates/ must be syntactically valid Python code.
3. Style Consistency: All Python and Markdown files must be formatted according to the local style guides.

Running Validation Locally

You can run the validation checks locally before committing changes:

python3 run_presubmit.py

This runs the checks defined in PRESUBMIT.py using a mock harness.

Files in this Directory

• SKILL.md: The core protocol defining the creation stages and rules.
• README.md: This file, documenting usage and verification.
• PRESUBMIT.py: Static checks for link integrity and template syntax.
• run_presubmit.py: Local runner for presubmit checks.
• .style.yapf: Python style configuration (PEP8).
• .style.mdformat: Markdown style marker.
• templates/: Templates used by the skill to generate new workflows.