$ show msitarzewski/agency-agents
This repository is a curated library of over 230 specialized AI agent definitions, each with a detailed persona, mission, rules, deliverables, and workflow. It serves as a multi-agent orchestration framework (NEXUS) that coordinates agents through phases with quality gates and handoffs. The agents are designed to be installed into various AI coding tools (Claude Code, Cursor, Copilot, etc.) via a modular installer and conversion pipeline. The repository also includes strategy documents, runbooks, and coordination templates for managing multi-agent projects. While primarily a collection of prompt definitions, it is supported by a robust set of scripts for validation, installation, and consistency checking, making it a practical tool for deploying AI agents in real-world workflows.
The LLM identified 5 files worth highlighting for reuse:
Shared bash library providing functions for parsing YAML frontmatter from markdown files, generating slugs, and implementing a terminal UI wizard with alt-screen, keyboard navigation, and flicker-free drawing. This is a reusable pattern for any bash-based tool that needs to parse structured metadata or present an interactive selection menu.
Interactive CLI installer that supports installing agents into 14+ AI coding tools (Claude Code, Cursor, Copilot, etc.) with options for division, agent, or file-based selection. Demonstrates a modular installer pattern where each tool has a dedicated function, and the selection engine supports filtering and parallel installation. Reusable for any project that needs to deploy configuration files to multiple target platforms.
Converts agent markdown files into 14 tool-specific formats (TOML, YAML, .mdc, SKILL.md, etc.) using a single source of truth. Demonstrates a clean pipeline pattern: read once, write to multiple formats, with support for parallel conversion and tool-specific renderers. Reusable for any project that needs to generate multiple output formats from a canonical source.
Defines 7 standardized handoff document templates for agent-to-agent communication, including standard handoff, QA pass/fail, escalation, phase gate, sprint review, and incident handoff. This is a reusable pattern for any multi-agent system that needs structured, traceable transitions between agents or phases.
Reference implementation of a multi-agent workflow that uses an MCP memory server for persistent cross-session context. Demonstrates a pattern for agents to store and recall information using `remember`, `recall`, and `rollback` operations with project tags. Reusable for any agent system that needs to maintain state across interactions.