Analysis updated 2026-05-18
Load a specialist agent persona like a security engineer or backend architect into your AI coding tool.
Auto-install agent files for whichever AI tools you already have set up.
Switch between prototyping and code-review personas without rewriting prompts each time.
| ucoingroup/agency-agents | anthonyhann/knowledge-wiki | baiyuetribe/test-heroku | |
|---|---|---|---|
| Stars | 1 | 1 | 1 |
| Language | Shell | Shell | Shell |
| Last pushed | — | — | 2021-06-30 |
| Maintenance | — | — | Dormant |
| Setup difficulty | easy | moderate | hard |
| Complexity | 1/5 | 3/5 | 1/5 |
| Audience | developer | developer | developer |
Figures from each repo's GitHub metadata at analysis time.
A shell script auto-detects installed AI tools and places the agent files automatically.
The Agency is a collection of AI agent personality files, detailed prompt instructions that give AI coding assistants a specific role, expertise, and working style. Rather than using a generic "help me code" prompt, you load one of these agents to get a specialist: a frontend developer focused on building user interfaces, a backend architect who thinks about scalability and APIs, a security engineer who looks for vulnerabilities, a DevOps automator who handles deployment pipelines, and many more. Each agent comes with a defined personality, a set of workflows it follows, and an emphasis on producing real, usable output. The collection works with a range of AI tools including Claude Code, GitHub Copilot, Gemini CLI, Cursor, Aider, Windsurf, and others. Installation is handled by a shell script that auto-detects which tools you have installed and places the agent files in the right location. A separate convert script generates integration files tailored to each tool's format. The engineering division listed in the README includes agents for frontend development, backend architecture, mobile apps, AI integration, DevOps automation, security, embedded firmware, incident response, database optimization, and several others. The README is truncated so the full roster is longer. It is most useful for developers who want to quickly switch context, activating the "Rapid Prototyper" when building an MVP, then the "Code Reviewer" before a pull request, without crafting detailed instructions from scratch each time. The project is MIT licensed and written primarily as Shell scripts with agent definitions stored as markdown files.
A collection of AI agent personality prompt files that turn coding assistants like Claude Code or Cursor into role-specific specialists.
Mainly Shell. The stack also includes Shell, Markdown.
Use freely for any purpose, including commercial use, as long as you keep the copyright notice.
Setup difficulty is rated easy, with roughly 5min to a first successful run.
Mainly developer.
This repo across BitVibe Labs
Verify against the repo before relying on details.