conorbronsdon

Agent Skill Builder — Testing skill for Claude Code

Testing community

A skill that builds agent skills — and doesn't rot.

How to install Agent Skill Builder

This entry records only its repository, not the path inside it, so there is no exact command to give. Open conorbronsdon/agent-skill-builder and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Agent Skill Builder does

A skill that builds agent skills — and doesn't rot. Spec-current frontmatter plus a validator that fails on drift.

Alternatives in Testing

README

agent-skill-builder

A Claude Code-first skill builder with portable Agent Skills validation and drift detection that can fail.

[![GitHub stars](https://img.shields.io/github/stars/conorbronsdon/agent-skill-builder?style=social)](https://github.com/conorbronsdon/agent-skill-builder/stargazers) [![License: MIT](https://img.shields.io/badge/License-MIT-blue.svg?style=flat-square)](LICENSE) [![Podcast](https://img.shields.io/badge/Podcast-Chain_of_Thought-purple?style=flat-square)](https://chainofthought.show/?utm_source=github&utm_medium=referral&utm_campaign=repo-readme&utm_content=agent-skill-builder) [![X](https://img.shields.io/badge/X-@ConorBronsdon-black?style=flat-square&logo=x)](https://x.com/ConorBronsdon)

Terminal demo showing the validator pass valid skills and fail a deliberately broken one.


Skill generators age fast. Frontmatter changes, runtime-specific fields leak into portable skills, and structural validators pass because nobody tested whether they can fail.

`agent-skill-builder` handles those as explicit design decisions. It builds Claude Code skills by default, can target the portable [Agent Skills](https://agentskills.io) core, validates each runtime separately, and checks the live Claude Code field set every week.

What it does

Describe the skill you want in plain language. The builder decides:

  • Runtime: Claude Code extensions or the portable Agent Skills core.
  • Skill shape: reference, bounded task, or tool-backed workflow.
  • Invocation control: user-only, model-only, or discoverable both ways.
  • Arguments: $ARGUMENTS, positional $N, or named arguments wired into the body.
  • Context budget: trigger language in the description, conditional detail in references.
  • Execution safety: scoped tool grants, explicit mutation gates, and read-only dynamic injection.

Three modes cover the lifecycle:

/agent-skill-builder new