awesome-codex-subagents
Compare original and translation side by side
🇺🇸
Original
English🇨🇳
Translation
Chineseawesome-codex-subagents
awesome-codex-subagents
Skill by ara.so — Codex Skills collection.
A definitive collection of 136+ specialized Codex subagents covering development, infrastructure, quality assurance, and domain-specific tasks. Each subagent is a configuration file that extends Codex with focused expertise for specific development scenarios.
.toml由ara.so开发的Skill — Codex Skills合集。
这是一个包含136+个专用Codex子代理的权威合集,涵盖开发、基础设施、质量保障以及特定领域任务。每个子代理都是一个配置文件,可为Codex扩展针对特定开发场景的专业能力。
.tomlWhat This Project Does
本项目功能
This repository provides pre-configured Codex subagents that you can install to delegate specialized tasks to AI assistants with domain expertise. Instead of asking a general-purpose agent to handle everything, you can explicitly invoke subagents optimized for:
- Core Development: API design, frontend/backend development, fullstack work
- Language Specialists: Python, TypeScript, Go, Rust, Java, and 20+ more languages
- Infrastructure: DevOps, Kubernetes, Terraform, cloud architecture
- Quality & Security: Testing, security audits, accessibility, code review
- Data & Analytics: Data engineering, ML pipelines, analytics
- Content & Documentation: Technical writing, API docs, localization
- Domain-Specific: Fintech, healthcare, gaming, embedded systems
- Emerging Tech: Blockchain, AI/ML infrastructure, quantum computing
- Research & Tools: Web search, academic research, benchmarking
本仓库提供预配置的Codex子代理,您可以安装这些代理,将专业任务委托给具备领域知识的AI助手。无需让通用型代理处理所有事务,您可以明确调用针对以下场景优化的子代理:
- 核心开发:API设计、前端/后端开发、全栈开发
- 语言专家:Python、TypeScript、Go、Rust、Java及20余种其他语言
- 基础设施:DevOps、Kubernetes、Terraform、云架构
- 质量与安全:测试、安全审计、无障碍适配、代码评审
- 数据与分析:数据工程、ML流水线、数据分析
- 内容与文档:技术写作、API文档、本地化
- 特定领域:金融科技、医疗健康、游戏开发、嵌入式系统
- 新兴技术:区块链、AI/ML基础设施、量子计算
- 研究与工具:网页搜索、学术研究、基准测试
Installation
安装方法
Global Installation (Available in All Projects)
全局安装(所有项目可用)
bash
undefinedbash
undefinedClone the repository
Clone the repository
git clone https://github.com/VoltAgent/awesome-codex-subagents.git
cd awesome-codex-subagents
git clone https://github.com/VoltAgent/awesome-codex-subagents.git
cd awesome-codex-subagents
Create global agents directory
Create global agents directory
mkdir -p ~/.codex/agents
mkdir -p ~/.codex/agents
Install specific subagents (examples)
Install specific subagents (examples)
cp categories/01-core-development/backend-developer.toml ~/.codex/agents/
cp categories/02-language-specialists/python-pro.toml ~/.codex/agents/
cp categories/03-infrastructure/kubernetes-specialist.toml ~/.codex/agents/
undefinedcp categories/01-core-development/backend-developer.toml ~/.codex/agents/
cp categories/02-language-specialists/python-pro.toml ~/.codex/agents/
cp categories/03-infrastructure/kubernetes-specialist.toml ~/.codex/agents/
undefinedProject-Specific Installation (Higher Precedence)
项目专属安装(优先级更高)
bash
undefinedbash
undefinedIn your project root
In your project root
mkdir -p .codex/agents
mkdir -p .codex/agents
Install project-specific agents
Install project-specific agents
cp path/to/awesome-codex-subagents/categories/04-quality-security/reviewer.toml .codex/agents/
cp path/to/awesome-codex-subagents/categories/02-language-specialists/typescript-pro.toml .codex/agents/
undefinedcp path/to/awesome-codex-subagents/categories/04-quality-security/reviewer.toml .codex/agents/
cp path/to/awesome-codex-subagents/categories/02-language-specialists/typescript-pro.toml .codex/agents/
undefinedInstall All Subagents in a Category
安装某一分类下的所有子代理
bash
undefinedbash
undefinedInstall all language specialists globally
Install all language specialists globally
mkdir -p ~/.codex/agents
cp categories/02-language-specialists/*.toml ~/.codex/agents/
mkdir -p ~/.codex/agents
cp categories/02-language-specialists/*.toml ~/.codex/agents/
Install all infrastructure agents for current project
Install all infrastructure agents for current project
mkdir -p .codex/agents
cp categories/03-infrastructure/*.toml .codex/agents/
undefinedmkdir -p .codex/agents
cp categories/03-infrastructure/*.toml .codex/agents/
undefinedSubagent Structure
子代理结构
Each subagent is a file with this structure:
.tomltoml
name = "python-pro"
description = "Python ecosystem master for development, testing, and packaging"
model = "gpt-5.3-codex-spark"
model_reasoning_effort = "medium"
sandbox_mode = "workspace-write"
[instructions]
text = """
You are a Python development expert specializing in modern Python 3.10+...
Core Responsibilities:
- Write idiomatic, type-hinted Python code
- Use virtual environments and modern packaging tools
- Implement comprehensive testing with pytest
...
"""每个子代理都是一个文件,结构如下:
.tomltoml
name = "python-pro"
description = "Python ecosystem master for development, testing, and packaging"
model = "gpt-5.3-codex-spark"
model_reasoning_effort = "medium"
sandbox_mode = "workspace-write"
[instructions]
text = """
You are a Python development expert specializing in modern Python 3.10+...
Core Responsibilities:
- Write idiomatic, type-hinted Python code
- Use virtual environments and modern packaging tools
- Implement comprehensive testing with pytest
...
"""Key Configuration Fields
关键配置字段
- name: Unique identifier for the subagent
- description: When to invoke this subagent (used by Codex for routing)
- model: Which GPT model to use (for deep reasoning,
gpt-5.4for fast tasks)gpt-5.3-codex-spark - sandbox_mode: Filesystem access (,
read-only, orworkspace-write)full - instructions.text: The system prompt defining the subagent's expertise and behavior
- name:子代理的唯一标识符
- description:调用该子代理的场景(Codex用于路由)
- model:使用的GPT模型(深度推理用,快速任务用
gpt-5.4)gpt-5.3-codex-spark - sandbox_mode:文件系统权限(、
read-only或workspace-write)full - instructions.text:定义子代理专业能力与行为的系统提示词
Using Subagents
使用子代理
Explicit Delegation in Prompts
在提示词中明确委托
Codex does not auto-spawn custom subagents. You must explicitly delegate:
bash
undefinedCodex不会自动生成自定义子代理,您必须明确委托:
bash
undefinedInvoke the backend developer subagent
Invoke the backend developer subagent
"@backend-developer create a REST API for user management with FastAPI"
"@backend-developer create a REST API for user management with FastAPI"
Use the security auditor to review code
Use the security auditor to review code
"@security-auditor review the authentication module for vulnerabilities"
"@security-auditor review the authentication module for vulnerabilities"
Get the Python specialist to refactor code
Get the Python specialist to refactor code
"@python-pro refactor this script to use type hints and dataclasses"
undefined"@python-pro refactor this script to use type hints and dataclasses"
undefinedMultiple Subagents in Sequence
按顺序调用多个子代理
bash
undefinedbash
undefinedDesign API first, then implement
Design API first, then implement
"@api-designer design a RESTful API for a blog system"
"@api-designer design a RESTful API for a blog system"
After reviewing the design:
After reviewing the design:
"@backend-developer implement the blog API using the design from api-designer"
undefined"@backend-developer implement the blog API using the design from api-designer"
undefinedProject-Specific Overrides
项目专属覆盖
If both global and project-specific agents exist with the same name, the project-specific one takes precedence:
bash
undefined如果全局和项目专属代理存在同名情况,项目专属代理优先级更高:
bash
undefinedGlobal agent at ~/.codex/agents/reviewer.toml
Global agent at ~/.codex/agents/reviewer.toml
Project agent at .codex/agents/reviewer.toml
Project agent at .codex/agents/reviewer.toml
The project version will be used when you invoke @reviewer
The project version will be used when you invoke @reviewer
undefinedundefinedCommon Patterns
常见模式
Backend Development Workflow
后端开发工作流
bash
undefinedbash
undefined1. Design the API
1. Design the API
mkdir -p .codex/agents
cp categories/01-core-development/api-designer.toml .codex/agents/
mkdir -p .codex/agents
cp categories/01-core-development/api-designer.toml .codex/agents/
2. Install language-specific agent
2. Install language-specific agent
cp categories/02-language-specialists/python-pro.toml .codex/agents/
cp categories/02-language-specialists/python-pro.toml .codex/agents/
3. Add security review
3. Add security review
cp categories/04-quality-security/security-auditor.toml .codex/agents/
cp categories/04-quality-security/security-auditor.toml .codex/agents/
Usage:
Usage:
"@api-designer design a REST API for inventory management"
"@python-pro implement the inventory API with FastAPI and SQLAlchemy"
"@security-auditor review the authentication and authorization logic"
undefined"@api-designer design a REST API for inventory management"
"@python-pro implement the inventory API with FastAPI and SQLAlchemy"
"@security-auditor review the authentication and authorization logic"
undefinedInfrastructure Setup
基础设施搭建
bash
undefinedbash
undefinedInstall infrastructure agents
Install infrastructure agents
mkdir -p ~/.codex/agents
cp categories/03-infrastructure/terraform-engineer.toml ~/.codex/agents/
cp categories/03-infrastructure/kubernetes-specialist.toml ~/.codex/agents/
cp categories/03-infrastructure/cloud-architect.toml ~/.codex/agents/
mkdir -p ~/.codex/agents
cp categories/03-infrastructure/terraform-engineer.toml ~/.codex/agents/
cp categories/03-infrastructure/kubernetes-specialist.toml ~/.codex/agents/
cp categories/03-infrastructure/cloud-architect.toml ~/.codex/agents/
Usage:
Usage:
"@cloud-architect design AWS infrastructure for a multi-region web app"
"@terraform-engineer write Terraform modules for the AWS design"
"@kubernetes-specialist create Kubernetes manifests for the application"
undefined"@cloud-architect design AWS infrastructure for a multi-region web app"
"@terraform-engineer write Terraform modules for the AWS design"
"@kubernetes-specialist create Kubernetes manifests for the application"
undefinedFull-Stack Feature Development
全栈功能开发
bash
undefinedbash
undefinedInstall full-stack agents
Install full-stack agents
cp categories/02-language-specialists/react-specialist.toml .codex/agents/
cp categories/02-language-specialists/nodejs-expert.toml .codex/agents/
cp categories/04-quality-security/e2e-tester.toml .codex/agents/
cp categories/02-language-specialists/react-specialist.toml .codex/agents/
cp categories/02-language-specialists/nodejs-expert.toml .codex/agents/
cp categories/04-quality-security/e2e-tester.toml .codex/agents/
Usage:
Usage:
"@react-specialist build a product listing page with filtering and pagination"
"@nodejs-expert create Express API endpoints for product data"
"@e2e-tester write Playwright tests for the product listing flow"
undefined"@react-specialist build a product listing page with filtering and pagination"
"@nodejs-expert create Express API endpoints for product data"
"@e2e-tester write Playwright tests for the product listing flow"
undefinedCode Quality Pipeline
代码质量流水线
bash
undefinedbash
undefinedInstall quality agents
Install quality agents
cp categories/04-quality-security/reviewer.toml .codex/agents/
cp categories/04-quality-security/test-engineer.toml .codex/agents/
cp categories/04-quality-security/accessibility-tester.toml .codex/agents/
cp categories/04-quality-security/reviewer.toml .codex/agents/
cp categories/04-quality-security/test-engineer.toml .codex/agents/
cp categories/04-quality-security/accessibility-tester.toml .codex/agents/
Usage:
Usage:
"@reviewer analyze the new payment module for design issues"
"@test-engineer add unit and integration tests for the payment module"
"@accessibility-tester audit the checkout page for WCAG 2.1 AA compliance"
undefined"@reviewer analyze the new payment module for design issues"
"@test-engineer add unit and integration tests for the payment module"
"@accessibility-tester audit the checkout page for WCAG 2.1 AA compliance"
undefinedConfiguration
配置
Custom Subagent Configuration
自定义子代理配置
Create in your project:
.codex/config.tomltoml
[agents]在项目中创建:
.codex/config.tomltoml
[agents]Override default models for specific agents
Override default models for specific agents
python-pro.model = "gpt-5.4" # Use more powerful model
reviewer.sandbox_mode = "read-only" # Restrict to read-only
python-pro.model = "gpt-5.4" # Use more powerful model
reviewer.sandbox_mode = "read-only" # Restrict to read-only
Set default reasoning effort
Set default reasoning effort
*.model_reasoning_effort = "high"
undefined*.model_reasoning_effort = "high"
undefinedCreating Custom Subagents
创建自定义子代理
Create a new file in :
.toml.codex/agents/toml
name = "my-custom-agent"
description = "Specialized agent for my team's specific needs"
model = "gpt-5.3-codex-spark"
sandbox_mode = "workspace-write"
[instructions]
text = """
You are a custom development agent for [YOUR TEAM/PROJECT].
Your primary responsibilities:
1. Follow our team's coding standards (link to internal docs)
2. Use our specific tech stack: [list technologies]
3. Implement features according to our architecture patterns
Code Style:
- Use [specific linter/formatter]
- Follow [naming conventions]
- Include [specific testing patterns]
Always:
- Check our internal documentation at [URL]
- Reference our API patterns in [repo location]
- Use our shared component library [package name]
"""在中创建新的文件:
.codex/agents/.tomltoml
name = "my-custom-agent"
description = "Specialized agent for my team's specific needs"
model = "gpt-5.3-codex-spark"
sandbox_mode = "workspace-write"
[instructions]
text = """
You are a custom development agent for [YOUR TEAM/PROJECT].
Your primary responsibilities:
1. Follow our team's coding standards (link to internal docs)
2. Use our specific tech stack: [list technologies]
3. Implement features according to our architecture patterns
Code Style:
- Use [specific linter/formatter]
- Follow [naming conventions]
- Include [specific testing patterns]
Always:
- Check our internal documentation at [URL]
- Reference our API patterns in [repo location]
- Use our shared component library [package name]
"""Category Overview
分类概览
Core Development (12 agents)
核心开发(12个代理)
- - REST and GraphQL API design
api-designer - - Server-side development
backend-developer - - UI/UX implementation
frontend-developer - - End-to-end features
fullstack-developer - - Cross-platform mobile apps
mobile-developer
- - REST和GraphQL API设计
api-designer - - 服务端开发
backend-developer - - UI/UX实现
frontend-developer - - 端到端功能开发
fullstack-developer - - 跨平台移动应用开发
mobile-developer
Language Specialists (28 agents)
语言专家(28个代理)
- - Python ecosystem expert
python-pro - - TypeScript development
typescript-pro - - Systems programming
rust-engineer - - Go concurrency and services
golang-pro - - React 18+ patterns
react-specialist - - Next.js 14+ full-stack
nextjs-developer - - Vue 3 Composition API
vue-expert - - Angular 15+ enterprise
angular-architect
- - Python生态专家
python-pro - - TypeScript开发
typescript-pro - - 系统编程
rust-engineer - - Go并发与服务开发
golang-pro - - React 18+模式
react-specialist - - Next.js 14+全栈开发
nextjs-developer - - Vue 3组合式API
vue-expert - - Angular 15+企业级开发
angular-architect
Infrastructure (16 agents)
基础设施(16个代理)
- - CI/CD pipelines
devops-engineer - - K8s orchestration
kubernetes-specialist - - Infrastructure as Code
terraform-engineer - - AWS/GCP/Azure design
cloud-architect - - Site reliability
sre-engineer - - Container optimization
docker-expert
- - CI/CD流水线
devops-engineer - - K8s编排
kubernetes-specialist - - 基础设施即代码
terraform-engineer - - AWS/GCP/Azure架构设计
cloud-architect - - 站点可靠性工程
sre-engineer - - 容器优化
docker-expert
Quality & Security (16 agents)
质量与安全(16个代理)
- - Vulnerability assessment
security-auditor - - Testing strategy
test-engineer - - Code review specialist
reviewer - - WCAG compliance
accessibility-tester - - Performance tuning
performance-optimizer
- - 漏洞评估
security-auditor - - 测试策略
test-engineer - - 代码评审专家
reviewer - - WCAG合规性测试
accessibility-tester - - 性能调优
performance-optimizer
Troubleshooting
故障排查
Subagent Not Found
子代理未找到
bash
undefinedbash
undefinedCheck if subagent is installed
Check if subagent is installed
ls -la ~/.codex/agents/
ls -la .codex/agents/
ls -la ~/.codex/agents/
ls -la .codex/agents/
Verify the name matches the file
Verify the name matches the file
cat .codex/agents/python-pro.toml | grep "^name"
cat .codex/agents/python-pro.toml | grep "^name"
Restart Codex session
Restart Codex session
(Implementation-specific, usually closing and reopening)
(Implementation-specific, usually closing and reopening)
undefinedundefinedWrong Subagent Responding
错误的子代理响应
bash
undefinedbash
undefinedCheck for name conflicts
Check for name conflicts
find ~/.codex/agents .codex/agents -name "*.toml" -exec grep "^name" {} ; -print
find ~/.codex/agents .codex/agents -name "*.toml" -exec grep "^name" {} ; -print
Project-specific agents override global ones
Project-specific agents override global ones
Remove the duplicate or rename one:
Remove the duplicate or rename one:
mv .codex/agents/python-pro.toml .codex/agents/python-pro-custom.toml
undefinedmv .codex/agents/python-pro.toml .codex/agents/python-pro-custom.toml
undefinedSubagent Not Following Instructions
子代理未遵循指令
bash
undefinedbash
undefinedReview the instruction text
Review the instruction text
cat .codex/agents/your-agent.toml
cat .codex/agents/your-agent.toml
Ensure description clearly states when to invoke
Ensure description clearly states when to invoke
Update the instructions section to be more specific:
Update the instructions section to be more specific:
nano .codex/agents/your-agent.toml
nano .codex/agents/your-agent.toml
Be explicit in your delegation:
Be explicit in your delegation:
"@your-agent [very specific task description]"
undefined"@your-agent [very specific task description]"
undefinedPerformance Issues
性能问题
bash
undefinedbash
undefinedCheck model configuration
Check model configuration
cat .codex/agents/slow-agent.toml | grep "^model"
cat .codex/agents/slow-agent.toml | grep "^model"
Switch to faster model for simple tasks:
Switch to faster model for simple tasks:
Change model = "gpt-5.4" to model = "gpt-5.3-codex-spark"
Change model = "gpt-5.4" to model = "gpt-5.3-codex-spark"
Reduce reasoning effort in config:
Reduce reasoning effort in config:
[agents]
slow-agent.model_reasoning_effort = "low"
undefined[agents]
slow-agent.model_reasoning_effort = "low"
undefinedSandbox Restrictions
沙箱限制
bash
undefinedbash
undefinedIf agent can't modify files:
If agent can't modify files:
Check sandbox_mode in the .toml file
Check sandbox_mode in the .toml file
cat .codex/agents/your-agent.toml | grep "sandbox_mode"
cat .codex/agents/your-agent.toml | grep "sandbox_mode"
Update to allow writes:
Update to allow writes:
sandbox_mode = "workspace-write"
sandbox_mode = "workspace-write"
Or for full system access (use cautiously):
Or for full system access (use cautiously):
sandbox_mode = "full"
undefinedsandbox_mode = "full"
undefinedReal-World Examples
实际案例
Example 1: Building a REST API with Python
案例1:使用Python构建REST API
bash
undefinedbash
undefinedInstall agents
Install agents
cp categories/02-language-specialists/python-pro.toml .codex/agents/
cp categories/01-core-development/api-designer.toml .codex/agents/
cp categories/02-language-specialists/python-pro.toml .codex/agents/
cp categories/01-core-development/api-designer.toml .codex/agents/
Step 1: Design
Step 1: Design
"@api-designer design a REST API for a task management system with users, projects, and tasks"
"@api-designer design a REST API for a task management system with users, projects, and tasks"
Step 2: Implement
Step 2: Implement
"@python-pro implement the task management API using FastAPI with:
- JWT authentication
- SQLAlchemy models
- Pydantic schemas
- CRUD endpoints for users, projects, and tasks
- PostgreSQL database"
"@python-pro implement the task management API using FastAPI with:
- JWT authentication
- SQLAlchemy models
- Pydantic schemas
- CRUD endpoints for users, projects, and tasks
- PostgreSQL database"
Step 3: Test
Step 3: Test
"@python-pro add pytest tests for all endpoints with fixtures and mocks"
undefined"@python-pro add pytest tests for all endpoints with fixtures and mocks"
undefinedExample 2: Infrastructure with Terraform
案例2:使用Terraform搭建基础设施
bash
undefinedbash
undefinedInstall agents
Install agents
cp categories/03-infrastructure/terraform-engineer.toml .codex/agents/
cp categories/03-infrastructure/cloud-architect.toml .codex/agents/
cp categories/03-infrastructure/terraform-engineer.toml .codex/agents/
cp categories/03-infrastructure/cloud-architect.toml .codex/agents/
Design infrastructure
Design infrastructure
"@cloud-architect design AWS infrastructure for a containerized web application with:
- ECS Fargate for compute
- RDS PostgreSQL for database
- ElastiCache Redis for sessions
- ALB for load balancing
- S3 for static assets
- CloudFront for CDN"
"@cloud-architect design AWS infrastructure for a containerized web application with:
- ECS Fargate for compute
- RDS PostgreSQL for database
- ElastiCache Redis for sessions
- ALB for load balancing
- S3 for static assets
- CloudFront for CDN"
Implement with Terraform
Implement with Terraform
"@terraform-engineer create Terraform modules for the AWS infrastructure design:
- Use remote state in S3
- Separate modules for VPC, ECS, RDS, Redis, ALB
- Variables for environment-specific config
- Outputs for endpoints and connection strings"
undefined"@terraform-engineer create Terraform modules for the AWS infrastructure design:
- Use remote state in S3
- Separate modules for VPC, ECS, RDS, Redis, ALB
- Variables for environment-specific config
- Outputs for endpoints and connection strings"
undefinedExample 3: Frontend Development with React
案例3:使用React进行前端开发
bash
undefinedbash
undefinedInstall agents
Install agents
cp categories/02-language-specialists/react-specialist.toml .codex/agents/
cp categories/02-language-specialists/typescript-pro.toml .codex/agents/
cp categories/02-language-specialists/react-specialist.toml .codex/agents/
cp categories/02-language-specialists/typescript-pro.toml .codex/agents/
Build UI component
Build UI component
"@react-specialist create a DataTable component with:
- TypeScript types
- Column sorting
- Pagination
- Row selection
- Filtering
- Virtualized scrolling for large datasets
- Tailwind CSS styling"
"@react-specialist create a DataTable component with:
- TypeScript types
- Column sorting
- Pagination
- Row selection
- Filtering
- Virtualized scrolling for large datasets
- Tailwind CSS styling"
Add tests
Add tests
"@react-specialist write React Testing Library tests for DataTable covering:
- Rendering with mock data
- Sorting functionality
- Pagination controls
- Filter interactions"
undefined"@react-specialist write React Testing Library tests for DataTable covering:
- Rendering with mock data
- Sorting functionality
- Pagination controls
- Filter interactions"
undefinedExample 4: Security Review Pipeline
案例4:安全评审流水线
bash
undefinedbash
undefinedInstall security agents
Install security agents
cp categories/04-quality-security/security-auditor.toml .codex/agents/
cp categories/04-quality-security/dependency-auditor.toml .codex/agents/
cp categories/04-quality-security/security-auditor.toml .codex/agents/
cp categories/04-quality-security/dependency-auditor.toml .codex/agents/
Audit authentication code
Audit authentication code
"@security-auditor review the authentication system in src/auth/ for:
- SQL injection vulnerabilities
- XSS risks
- CSRF protection
- Session management
- Password hashing
- Rate limiting"
"@security-auditor review the authentication system in src/auth/ for:
- SQL injection vulnerabilities
- XSS risks
- CSRF protection
- Session management
- Password hashing
- Rate limiting"
Check dependencies
Check dependencies
"@dependency-auditor scan package.json and identify:
- Known CVEs in dependencies
- Outdated packages with security patches
- License compliance issues
- Recommended updates"
undefined"@dependency-auditor scan package.json and identify:
- Known CVEs in dependencies
- Outdated packages with security patches
- License compliance issues
- Recommended updates"
undefinedEnvironment Variables
环境变量
Subagents should reference environment variables for sensitive data:
python
undefined子代理应引用环境变量存储敏感数据:
python
undefined✅ Correct - use environment variables
✅ Correct - use environment variables
import os
DATABASE_URL = os.getenv("DATABASE_URL")
API_KEY = os.getenv("API_KEY")
import os
DATABASE_URL = os.getenv("DATABASE_URL")
API_KEY = os.getenv("API_KEY")
❌ Incorrect - never hardcode secrets
❌ Incorrect - never hardcode secrets
DATABASE_URL = "postgresql://user:pass@localhost/db"
API_KEY = "sk-1234567890abcdef"
When instructing subagents:
```bash
"@python-pro create a database connection using the DATABASE_URL environment variable from .env"
"@nodejs-expert set up AWS SDK to use credentials from AWS_ACCESS_KEY_ID and AWS_SECRET_ACCESS_KEY environment variables"DATABASE_URL = "postgresql://user:pass@localhost/db"
API_KEY = "sk-1234567890abcdef"
在指令子代理时:
```bash
"@python-pro create a database connection using the DATABASE_URL environment variable from .env"
"@nodejs-expert set up AWS SDK to use credentials from AWS_ACCESS_KEY_ID and AWS_SECRET_ACCESS_KEY environment variables"Additional Resources
额外资源
- Official Codex Documentation
- GitHub Repository
- Discord Community
- Related Collections - Agent skills, Claude subagents, OpenClaw skills
Contributing
贡献指南
To add custom subagents to your local collection:
- Create a new file following the structure above
.toml - Place it in (project) or
.codex/agents/(global)~/.codex/agents/ - Reference the official documentation for advanced configuration options
To contribute to the upstream repository, submit a pull request with your subagent in the appropriate category folder.
如需向本地合集添加自定义子代理:
- 按照上述结构创建新的文件
.toml - 将其放置在(项目专属)或
.codex/agents/(全局)目录下~/.codex/agents/ - 参考官方文档获取高级配置选项
如需向上游仓库贡献,请提交Pull Request,将您的子代理放入对应分类文件夹中。