Does it support both Windows and Linux environments?

Yes, the skill is designed to provide executable commands for both PowerShell (primary) and Bash (fallback) to ensure cross-platform compatibility for validation steps.

How does this skill improve AI reliability?

It uses a six-core framework—Commands, Testing, Project Structure, Code Style, Git Workflow, and Boundaries—to ensure AI assistants have all the context they need to perform tasks without ambiguity.

What files does the Agent Creator skill modify?

This skill specifically targets agent definition files such as agents.md, .github/agents/*.md, and *.agent.md files. It does not modify application source code unless explicitly requested.

What are the safety boundaries used by this skill?

It utilizes a three-tier model: 'Always' for safe tasks, 'Ask first' for sensitive changes like adding dependencies, and 'Never' for destructive actions like force-pushing or deleting tests.

Agent Creator

Name: Agent Creator
Author: AndyElessar

byAndyElessar

0•

개발자 도구

Architects, reviews, and optimizes agent definition files to ensure clear, executable, and safe AI behavior within any repository.

The Agent Creator skill is a specialized tool for developers looking to define precise operational boundaries and personas for AI assistants. It automates the creation and auditing of agents.md and .agent.md files, converting vague requirements into structured operating manuals. By enforcing a standardized framework that includes executable commands, project-specific knowledge, and a three-tier safety model (Always/Ask first/Never), it ensures that AI agents follow local coding standards and git workflows while preventing destructive actions. This skill is essential for maintaining consistent AI behavior across complex development environments.

주요 기능

01Three-tier boundary modeling to enforce repository safety and constraints

020 GitHub stars

03Severity-based review of existing agent instructions for executability

04Standardized six-core section coverage for comprehensive agent behavior

05Multi-shell support providing both PowerShell and Bash command validation

06Automated drafting of agents.md and .github/agents/*.md files

사용 사례

01Hardening repository safety by establishing explicit 'Never' rules for AI actions

02Transforming generic prompt instructions into repository-aware operating manuals

03Defining custom personas for specialized roles like security-agent or docs-agent

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add andyelessar/skills agents-creator

For use in Claude.ai and ChatGPT