Claude Code's org chart: 30 hires, zero salaries

What it does

This repository is essentially a well-organized prompt library for Claude Code’s sub-agent feature. Each markdown file defines a specialized role—like backend-architect, tiktok-strategist, or whimsy-injector—with YAML frontmatter, a detailed system prompt, and usage examples. Drop them into Claude Code’s agents directory and the tool invokes the appropriate persona by name or context, turning a single coding session into a staffed studio.

The interesting bit

The project treats prompt engineering as organizational design. Every agent is built around a “6-day sprint” philosophy, and some trigger proactively—test-writer-fixer wakes up after you modify code, whimsy-injector appears after UI edits. It’s an org chart implemented as filesystem hierarchy.

Key highlights

• 30+ role-specific agents across 8 departments, from engineering to legal compliance
• Each agent is a structured markdown file with YAML frontmatter and a 500+ word system prompt
• Proactive triggers: specific agents auto-invoke based on context (code changes, feature flags, UI edits)
• Rigid templating standards with required examples, commentary, and success metrics
• Zero dependencies beyond Claude Code’s proprietary sub-agent directory

Caveats

• Pure prompt templates with no executable logic; value depends entirely on Claude Code’s underlying capabilities
• Tightly coupled to Anthropic’s proprietary sub-agent feature and directory structure
• README mandates extensive structure (500+ words, 4 examples) but provides no visibility into actual output quality or comparative benchmarks

Verdict

Worth browsing if you’re a Claude Code user trying to systematize a solo or small-team workflow. Skip it if you’re looking for a framework-agnostic agent orchestration layer or actual automation beyond text generation.