social-skills: Instructions file for Claude Code

AGENTS.md

Project Overview

This repository (tbhb/social-skills) is a collection of satirical personality-module "skills" for AI coding agents. Each skill teaches an agent a dysfunctional corporate behavior — passive-aggressive emails, weaponized empathy, process theater, etc. The project is comedic and deliberately absurd; the skills are meant to be funny, not genuinely useful.

Skills are organized into four plugins and distributed via the npx skills CLI. They are consumed by agent frameworks that support the SKILL.md format (YAML frontmatter + markdown instructions).

Repository Structure

.
├── plugins/                           # Each plugin is a self-contained bundle
│   └── <plugin-name>/
│       ├── .claude-plugin/
│       │   └── plugin.json            # Plugin metadata (name, version, description, author)
│       ├── skills/
│       │   └── <skill-name>/
│       │       └── SKILL.md           # Skill definition (frontmatter + instructions)
│       └── README.md                  # Plugin-level documentation and skill catalog
├── .agents/
│   └── skills/                        # Installed/symlinked copies of skills for local agent use
├── skills-lock.json                   # Lock file tracking installed skill hashes
├── README.md                          # Public-facing documentation and plugin catalog
├── LICENSE                            # CC0 1.0 Universal (public domain)
└── AGENTS.md                          # This file

Plugin Anatomy

Each plugin lives in plugins/<plugin-name>/ and follows the Claude Code plugin convention:

plugin.json

{
  "name": "<plugin-name>",
  "version": "1.0.0",
  "description": "A short description of the plugin's theme.",
  "author": {
    "name": "Tony Burns",
    "url": "https://github.com/tbhb"
  }
}

Skill Files

Every skill lives in plugins/<plugin-name>/skills/<skill-name>/SKILL.md and follows this structure:

---
name: <skill-name>
description: >-
  A concise description of when and how the agent should activate this skill.
  Written as an instruction to the agent, not a summary for humans.
---

# The "<Display Name>" <Optional Subtitle>

One-line thematic summary.

## System Instruction

A single paragraph giving the agent its persona, behavioral rules,
and constraints. This is the core prompt injection.

## Example Interaction

A **User:** / **Agent:** exchange demonstrating the skill in action.