Universal Agents standardizes AI agent configurations across coding assistants. Developers benefit from a shared AGENTS.md file for consistent agent setups. It connects to AI coding assistants like Claude, Codex, and Gemini, streamlining workflows.
git clone https://github.com/leochiu-a/universal-agents.gitUniversal Agents standardizes AI agent configurations across coding assistants. Developers benefit from a shared AGENTS.md file for consistent agent setups. It connects to AI coding assistants like Claude, Codex, and Gemini, streamlining workflows.
[{"step":"Create a project directory and initialize it with a basic structure (e.g., `/src`, `/docs`, `/tests`).","action":"Run `mkdir -p project/{src/{backend,frontend},docs,tests} && cd project` in your terminal."},{"step":"Draft the AGENTS.md file using the prompt template. Customize roles, permissions, and tools for your project.","action":"Use the provided prompt template to generate the file. Replace [PLACEHOLDERS] with your project-specific details."},{"step":"Save the file in the root directory of your project as `AGENTS.md`.","action":"Run `touch AGENTS.md` and paste the generated content. Ensure the file is committed to version control (e.g., Git)."},{"step":"Configure your AI assistant to load the AGENTS.md file. For Claude, add it to your `.claude/settings.json` under `agent_files`. For other assistants, check their documentation for agent configuration.","action":"Example for Claude: `{ \"agent_files\": [\"AGENTS.md\"] }` in `.claude/settings.json`."},{"step":"Test the configuration by delegating a task to a specific agent (e.g., 'Ask the backend-architect to design the user authentication API').","action":"Use your AI assistant's chat interface to send a task request referencing the agent role (e.g., '@backend-architect Please design the user authentication API')."}]
No install command available. Check the GitHub repository for manual installation instructions.
git clone https://github.com/leochiu-a/universal-agentsCopy the install command above and run it in your terminal.
Launch Claude Code, Cursor, or your preferred AI coding agent.
Use the prompt template or examples below to test the skill.
Adapt the skill to your specific use case and workflow.
Generate a standardized AGENTS.md configuration file for [PROJECT_NAME]. Include agent roles for [ROLE_1], [ROLE_2], and [ROLE_3]. Define their permissions, tools, and constraints. Format the file for compatibility with [AI_ASSISTANT_NAME] (e.g., Claude, Codex, or Gemini). Ensure the file follows the Universal Agents specification for cross-platform use.
```markdown # AGENTS.md - Project: E-Commerce API Redesign ## Overview This file standardizes agent configurations for the E-Commerce API Redesign project, ensuring consistency across [AI_ASSISTANT_NAME] setups. ## Agent Definitions ### 1. Backend Architect (backend-architect) - **Role**: Design and optimize API endpoints, database schemas, and microservices architecture. - **Permissions**: - Read/write access to `/src/backend` and `/docs/api-specs` - Execute shell commands for dependency installation and service startup - No access to `/src/frontend` or `/tests` - **Tools**: - PostgreSQL CLI for schema management - Swagger/OpenAPI for API documentation - Docker for container orchestration - **Constraints**: - Must use Node.js v18+ and TypeScript - Cannot modify `/src/frontend` files - Must generate OpenAPI specs for all new endpoints ### 2. Frontend Developer (frontend-dev) - **Role**: Implement user interfaces, components, and state management for the e-commerce platform. - **Permissions**: - Read/write access to `/src/frontend` and `/public` - Execute shell commands for dependency installation and build processes - No access to `/src/backend` or `/docs` - **Tools**: - React v18 with TypeScript - TailwindCSS for styling - Jest for unit testing - **Constraints**: - Must follow the design system in `/docs/design-system.md` - Cannot modify API endpoints or backend logic - Must ensure mobile responsiveness for all components ### 3. QA Engineer (qa-engineer) - **Role**: Write and execute test cases, identify bugs, and validate fixes. - **Permissions**: - Read access to all project files - Execute shell commands for test execution and reporting - Write access to `/tests` and `/docs/test-reports` - **Tools**: - Jest for unit and integration tests - Playwright for end-to-end testing - Postman for API testing - **Constraints**: - Must test all new features and bug fixes - Cannot modify production code directly - Must document all test cases in `/docs/test-cases.md` ## Usage Instructions 1. Place this file in the root directory of the project. 2. Configure your AI assistant to load this file at startup. 3. Use the agent roles to delegate tasks (e.g., 'Ask the backend-architect to design the product catalog API'). 4. Update this file whenever project requirements or constraints change. ```
Automated receipt scanning for expense management
IronCalc is a spreadsheet engine and ecosystem
Enterprise workflow automation and service management platform
Customer feedback management made simple
Complete help desk solution for growing teams
The AI automation platform built for everyone
Take a free 3-minute scan and get personalized AI skill recommendations.
Take free scan