[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"tool-VoltAgent--awesome-codex-subagents":3,"similar-VoltAgent--awesome-codex-subagents":50},{"id":4,"github_repo":5,"name":6,"description_en":7,"description_zh":8,"ai_summary_zh":9,"readme_en":10,"readme_zh":11,"quickstart_zh":12,"use_case_zh":13,"hero_image_url":14,"owner_login":15,"owner_name":15,"owner_avatar_url":16,"owner_bio":17,"owner_company":18,"owner_location":18,"owner_email":19,"owner_twitter":20,"owner_website":21,"owner_url":22,"languages":18,"stars":23,"forks":24,"last_commit_at":25,"license":26,"difficulty_score":27,"env_os":28,"env_gpu":29,"env_ram":29,"env_deps":30,"category_tags":34,"github_topics":37,"view_count":27,"oss_zip_url":18,"oss_zip_packed_at":18,"status":45,"created_at":46,"updated_at":47,"faqs":48,"releases":49},6825,"VoltAgent\u002Fawesome-codex-subagents","awesome-codex-subagents","A collection of 130+ specialized Codex subagents covering a wide range of development use cases.","awesome-codex-subagents 是一个专为 OpenAI Codex 打造的开源资源库，汇集了超过 136 个针对特定开发场景优化的子代理（Subagents）。它将复杂的软件开发工作流拆解为精细化的专业角色，涵盖核心开发、代码审查、安全审计、架构设计等十大类别，旨在解决通用 AI 助手在垂直领域深度不足、指令理解不够精准的问题。\n\n通过引入“专家分工”机制，开发者可以在处理具体任务时调用专属代理，例如让\"backend-developer\"专注后端逻辑，或指派\"security-auditor\"进行漏洞扫描，从而显著提升代码生成的质量与安全性。该项目特别适合需要高效利用 AI 辅助编程的软件工程师、技术团队及全栈开发者使用。\n\n其技术亮点在于采用了原生的 `.toml` 配置文件格式，安装简便且易于集成。用户只需将配置文件放入全局或项目目录，即可在 Codex 中灵活调用。更值得一提的是，每个子代理内置了智能模型路由功能，能根据任务复杂度（如深度推理或快速编码）自动匹配最合适的模型版本，在保障输出质量的同时有效平衡计算成本。无论是构建大型系统还是维护日常项目，awes","awesome-codex-subagents 是一个专为 OpenAI Codex 打造的开源资源库，汇集了超过 136 个针对特定开发场景优化的子代理（Subagents）。它将复杂的软件开发工作流拆解为精细化的专业角色，涵盖核心开发、代码审查、安全审计、架构设计等十大类别，旨在解决通用 AI 助手在垂直领域深度不足、指令理解不够精准的问题。\n\n通过引入“专家分工”机制，开发者可以在处理具体任务时调用专属代理，例如让\"backend-developer\"专注后端逻辑，或指派\"security-auditor\"进行漏洞扫描，从而显著提升代码生成的质量与安全性。该项目特别适合需要高效利用 AI 辅助编程的软件工程师、技术团队及全栈开发者使用。\n\n其技术亮点在于采用了原生的 `.toml` 配置文件格式，安装简便且易于集成。用户只需将配置文件放入全局或项目目录，即可在 Codex 中灵活调用。更值得一提的是，每个子代理内置了智能模型路由功能，能根据任务复杂度（如深度推理或快速编码）自动匹配最合适的模型版本，在保障输出质量的同时有效平衡计算成本。无论是构建大型系统还是维护日常项目，awesome-codex-subagents 都能帮助开发者将 Codex 的能力发挥到极致，实现更专业的自动化协作。","\u003Ca href=\"https:\u002F\u002Fgithub.com\u002FVoltAgent\u002Fvoltagent\">\n    \u003Cimg width=\"1500\" height=\"500\" alt=\"codex\" src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FVoltAgent_awesome-codex-subagents_readme_5328e2808a49.png\" \u002F>\n\u003C\u002Fa>\n\n\u003Cbr \u002F>\n\u003Cbr \u002F>\n\n\u003Cdiv align=\"center\">\n    \u003Cstrong>The awesome collection of 136+ Codex subagents across 10 categories.\u003C\u002Fstrong>\n    \u003Cbr \u002F>\n    \u003Cbr \u002F>\n\u003C\u002Fdiv>\n\n   \n\u003Cdiv align=\"center\">\n    \n[![Awesome](https:\u002F\u002Fawesome.re\u002Fbadge.svg)](https:\u002F\u002Fawesome.re)\n![Subagent Count](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fsubagents-136-blue?style=classic)\n[![Last Update](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Flast-commit\u002FVoltAgent\u002Fawesome-codex-subagents?label=Last%20update&style=classic)](https:\u002F\u002Fgithub.com\u002FVoltAgent\u002Fawesome-codex-subagents)\n[![Discord](https:\u002F\u002Fimg.shields.io\u002Fdiscord\u002F1361559153780195478.svg?label=&logo=discord&logoColor=ffffff&color=7389D8&labelColor=6A7EC2)](https:\u002F\u002Fs.voltagent.dev\u002Fdiscord)\n\n\u003Cbr \u002F>\n\n\n\u003Cdiv align=\"center\">\n    \u003Cstrong>More awesome collections for developers\u003C\u002Fstrong>\n    \u003Cbr \u002F>\n    \u003Cbr \u002F>\n\u003C\u002Fdiv>\n\n[![Agent Skills](https:\u002F\u002Fimg.shields.io\u002Fstatic\u002Fv1?label=%E2%9A%A1%20Agent&message=Skills%2012k&color=black&style=classic)](https:\u002F\u002Fgithub.com\u002FVoltAgent\u002Fawesome-agent-skills)\n[![Claude Code Subagents](https:\u002F\u002Fimg.shields.io\u002Fstatic\u002Fv1?label=Claude&message=Code%20Subagents%2014k&color=D97757&style=classic&logo=claude&logoColor=D97757)](https:\u002F\u002Fgithub.com\u002FVoltAgent\u002Fawesome-claude-code-subagents)\n[![OpenClaw Skills](https:\u002F\u002Fimg.shields.io\u002Fstatic\u002Fv1?label=%F0%9F%A6%9E%20OpenClaw&message=Skills%2040k&color=f53e36&style=classic)](https:\u002F\u002Fgithub.com\u002FVoltAgent\u002Fawesome-openclaw-skills)\n[![AI Agent Papers](https:\u002F\u002Fimg.shields.io\u002Fstatic\u002Fv1?label=arxiv&message=Agent%20Papers%20328&color=b31b1b&style=classic&logo=arxiv)](https:\u002F\u002Fgithub.com\u002FVoltAgent\u002Fawesome-ai-agent-papers)\n\u003C\u002Fdiv>\n\n\n# Awesome Codex Subagents\n\nThis repository serves as the definitive collection of [Codex Subagents](https:\u002F\u002Fdevelopers.openai.com\u002Fcodex\u002Fsubagents), specialized AI assistants designed for specific development tasks. Written specifically for Codex and aligned with the official docs.\n\n## Installation\n\nUse Codex custom agent directories exactly as documented:\n\n- `~\u002F.codex\u002Fagents\u002F` for global agents (available in all projects)\n- `.codex\u002Fagents\u002F` for project-specific agents (higher precedence in that repo)\n\n1. Clone this repository.\n2. Copy the `.toml` agent files you want into one of the directories above.\n3. Restart or refresh your Codex session if needed.\n4. Delegate explicitly in prompts (Codex does not auto-spawn custom subagents).\n\nExamples:\n```bash\nmkdir -p ~\u002F.codex\u002Fagents\ncp categories\u002F01-core-development\u002Fbackend-developer.toml ~\u002F.codex\u002Fagents\u002F\n```\n\n```bash\nmkdir -p .codex\u002Fagents\ncp categories\u002F04-quality-security\u002Freviewer.toml .codex\u002Fagents\u002F\n```\n\nIf you use agent configuration in Codex, keep it in `.codex\u002Fconfig.toml` under `[agents]` as described in the official docs.\n\n\n### Subagent Storage Locations\n\n| Type | Path | Availability | Precedence |\n|------|------|--------------|------------|\n| Project Subagents | `.codex\u002Fagents\u002F` | Current project only | Higher |\n| Global Subagents | `~\u002F.codex\u002Fagents\u002F` | All projects | Lower |\n\nNote: When naming conflicts occur, project-specific subagents override global ones.\n\n\n## Subagent Structure\n\nEach subagent uses a Codex-native `.toml` format:\n\n```toml\nname = \"subagent-name\"\ndescription = \"When this agent should be invoked\"\nmodel = \"gpt-5.3-codex-spark\"\nmodel_reasoning_effort = \"medium\"\nsandbox_mode = \"read-only\"\n\n[instructions]\ntext = \"\"\"\nYou are a [role description and expertise areas]...\n\n[Agent-specific checklists, patterns, and guidelines]...\n\"\"\"\n```\n\n### Smart Model Routing\n\nEach subagent includes a `model` field that automatically routes it to the right model -- balancing quality and cost:\n\n| Model | When It's Used | Examples |\n|-------|----------------|----------|\n| `gpt-5.4` | Deep reasoning -- architecture reviews, security audits, financial logic | `security-auditor`, `architect-reviewer`, `fintech-engineer` |\n| `gpt-5.3-codex-spark` | Fast scanning, synthesis, and lighter research tasks | `search-specialist`, `docs-researcher`, `agent-installer` |\n\n### Sandbox Mode Philosophy\n\nEach subagent's `sandbox_mode` field controls filesystem access:\n- **Read-only agents** (reviewers, auditors): `sandbox_mode = \"read-only\"` - analyze without modifying\n- **Workspace-write agents** (developers, engineers): `sandbox_mode = \"workspace-write\"` - create and modify files\n\n\n## Categories\n\n### [01. Core Development](categories\u002F01-core-development\u002F)\n\nEssential development subagents for everyday coding tasks.\n\n- [**api-designer**](categories\u002F01-core-development\u002Fapi-designer.toml) - REST and GraphQL API architect\n- [**backend-developer**](categories\u002F01-core-development\u002Fbackend-developer.toml) - Server-side expert for scalable APIs\n- [**code-mapper**](categories\u002F01-core-development\u002Fcode-mapper.toml) - Code path mapping and ownership boundary analysis\n- [**electron-pro**](categories\u002F01-core-development\u002Felectron-pro.toml) - Desktop application expert\n- [**frontend-developer**](categories\u002F01-core-development\u002Ffrontend-developer.toml) - UI\u002FUX specialist for React, Vue, and Angular\n- [**fullstack-developer**](categories\u002F01-core-development\u002Ffullstack-developer.toml) - End-to-end feature development\n- [**graphql-architect**](categories\u002F01-core-development\u002Fgraphql-architect.toml) - GraphQL schema and federation expert\n- [**microservices-architect**](categories\u002F01-core-development\u002Fmicroservices-architect.toml) - Distributed systems designer\n- [**mobile-developer**](categories\u002F01-core-development\u002Fmobile-developer.toml) - Cross-platform mobile specialist\n- [**ui-designer**](categories\u002F01-core-development\u002Fui-designer.toml) - Visual design and interaction specialist\n- [**ui-fixer**](categories\u002F01-core-development\u002Fui-fixer.toml) - Smallest safe patch for reproduced UI issues\n- [**websocket-engineer**](categories\u002F01-core-development\u002Fwebsocket-engineer.toml) - Real-time communication specialist\n\n### [02. Language Specialists](categories\u002F02-language-specialists\u002F)\n\nLanguage-specific experts with deep framework knowledge.\n- [**angular-architect**](categories\u002F02-language-specialists\u002Fangular-architect.toml) - Angular 15+ enterprise patterns expert\n- [**cpp-pro**](categories\u002F02-language-specialists\u002Fcpp-pro.toml) - C++ performance expert\n- [**csharp-developer**](categories\u002F02-language-specialists\u002Fcsharp-developer.toml) - .NET ecosystem specialist\n- [**django-developer**](categories\u002F02-language-specialists\u002Fdjango-developer.toml) - Django 4+ web development expert\n- [**dotnet-core-expert**](categories\u002F02-language-specialists\u002Fdotnet-core-expert.toml) - .NET 8 cross-platform specialist\n- [**dotnet-framework-4.8-expert**](categories\u002F02-language-specialists\u002Fdotnet-framework-4.8-expert.toml) - .NET Framework legacy enterprise specialist\n- [**elixir-expert**](categories\u002F02-language-specialists\u002Felixir-expert.toml) - Elixir and OTP fault-tolerant systems expert\n- [**erlang-expert**](categories\u002F02-language-specialists\u002Ferlang-expert.toml) - Erlang\u002FOTP and rebar3 engineering expert\n- [**flutter-expert**](categories\u002F02-language-specialists\u002Fflutter-expert.toml) - Flutter 3+ cross-platform mobile expert\n- [**golang-pro**](categories\u002F02-language-specialists\u002Fgolang-pro.toml) - Go concurrency specialist\n- [**java-architect**](categories\u002F02-language-specialists\u002Fjava-architect.toml) - Enterprise Java expert\n- [**javascript-pro**](categories\u002F02-language-specialists\u002Fjavascript-pro.toml) - JavaScript development expert\n- [**kotlin-specialist**](categories\u002F02-language-specialists\u002Fkotlin-specialist.toml) - Modern JVM language expert\n- [**laravel-specialist**](categories\u002F02-language-specialists\u002Flaravel-specialist.toml) - Laravel 10+ PHP framework expert\n- [**nextjs-developer**](categories\u002F02-language-specialists\u002Fnextjs-developer.toml) - Next.js 14+ full-stack specialist\n- [**php-pro**](categories\u002F02-language-specialists\u002Fphp-pro.toml) - PHP web development expert\n- [**powershell-5.1-expert**](categories\u002F02-language-specialists\u002Fpowershell-5.1-expert.toml) - Windows PowerShell 5.1 and full .NET Framework automation specialist\n- [**powershell-7-expert**](categories\u002F02-language-specialists\u002Fpowershell-7-expert.toml) - Cross-platform PowerShell 7+ automation and modern .NET specialist\n- [**python-pro**](categories\u002F02-language-specialists\u002Fpython-pro.toml) - Python ecosystem master\n- [**rails-expert**](categories\u002F02-language-specialists\u002Frails-expert.toml) - Rails 8.1 rapid development expert\n- [**react-specialist**](categories\u002F02-language-specialists\u002Freact-specialist.toml) - React 18+ modern patterns expert\n- [**rust-engineer**](categories\u002F02-language-specialists\u002Frust-engineer.toml) - Systems programming expert\n- [**spring-boot-engineer**](categories\u002F02-language-specialists\u002Fspring-boot-engineer.toml) - Spring Boot 3+ microservices expert\n- [**sql-pro**](categories\u002F02-language-specialists\u002Fsql-pro.toml) - Database query expert\n- [**swift-expert**](categories\u002F02-language-specialists\u002Fswift-expert.toml) - iOS and macOS specialist\n- [**typescript-pro**](categories\u002F02-language-specialists\u002Ftypescript-pro.toml) - TypeScript specialist\n- [**vue-expert**](categories\u002F02-language-specialists\u002Fvue-expert.toml) - Vue 3 Composition API expert\n\n\n### [03. Infrastructure](categories\u002F03-infrastructure\u002F)\n\nDevOps, cloud, and deployment specialists.\n\n- [**azure-infra-engineer**](categories\u002F03-infrastructure\u002Fazure-infra-engineer.toml) - Azure infrastructure and Az PowerShell automation expert\n- [**cloud-architect**](categories\u002F03-infrastructure\u002Fcloud-architect.toml) - AWS\u002FGCP\u002FAzure specialist\n- [**database-administrator**](categories\u002F03-infrastructure\u002Fdatabase-administrator.toml) - Database management expert\n- [**deployment-engineer**](categories\u002F03-infrastructure\u002Fdeployment-engineer.toml) - Deployment automation specialist\n- [**devops-engineer**](categories\u002F03-infrastructure\u002Fdevops-engineer.toml) - CI\u002FCD and automation expert\n- [**devops-incident-responder**](categories\u002F03-infrastructure\u002Fdevops-incident-responder.toml) - DevOps incident management\n- [**docker-expert**](categories\u002F03-infrastructure\u002Fdocker-expert.toml) - Docker containerization and optimization expert\n- [**incident-responder**](categories\u002F03-infrastructure\u002Fincident-responder.toml) - System incident response expert\n- [**kubernetes-specialist**](categories\u002F03-infrastructure\u002Fkubernetes-specialist.toml) - Container orchestration master\n- [**network-engineer**](categories\u002F03-infrastructure\u002Fnetwork-engineer.toml) - Network infrastructure specialist\n- [**platform-engineer**](categories\u002F03-infrastructure\u002Fplatform-engineer.toml) - Platform architecture expert\n- [**security-engineer**](categories\u002F03-infrastructure\u002Fsecurity-engineer.toml) - Infrastructure security specialist\n- [**sre-engineer**](categories\u002F03-infrastructure\u002Fsre-engineer.toml) - Site reliability engineering expert\n- [**terraform-engineer**](categories\u002F03-infrastructure\u002Fterraform-engineer.toml) - Infrastructure as Code expert\n- [**terragrunt-expert**](categories\u002F03-infrastructure\u002Fterragrunt-expert.toml) - Terragrunt orchestration and DRY IaC specialist\n- [**windows-infra-admin**](categories\u002F03-infrastructure\u002Fwindows-infra-admin.toml) - Active Directory, DNS, DHCP, and GPO automation specialist\n\n\u003Cdetails>\n\u003Csummary>\u003Cb>04. Quality & Security\u003C\u002Fb> — Testing, security, and code quality experts (16 agents)\u003C\u002Fsummary>\n\n### [04. Quality & Security](categories\u002F04-quality-security\u002F)\n\n- [**accessibility-tester**](categories\u002F04-quality-security\u002Faccessibility-tester.toml) - A11y compliance expert\n- [**ad-security-reviewer**](categories\u002F04-quality-security\u002Fad-security-reviewer.toml) - Active Directory security and GPO audit specialist\n- [**architect-reviewer**](categories\u002F04-quality-security\u002Farchitect-reviewer.toml) - Architecture review specialist\n- [**browser-debugger**](categories\u002F04-quality-security\u002Fbrowser-debugger.toml) - Browser-based reproduction and client-side debugging\n- [**chaos-engineer**](categories\u002F04-quality-security\u002Fchaos-engineer.toml) - System resilience testing expert\n- [**code-reviewer**](categories\u002F04-quality-security\u002Fcode-reviewer.toml) - Code quality guardian\n- [**compliance-auditor**](categories\u002F04-quality-security\u002Fcompliance-auditor.toml) - Regulatory compliance expert\n- [**debugger**](categories\u002F04-quality-security\u002Fdebugger.toml) - Advanced debugging specialist\n- [**error-detective**](categories\u002F04-quality-security\u002Ferror-detective.toml) - Error analysis and resolution expert\n- [**penetration-tester**](categories\u002F04-quality-security\u002Fpenetration-tester.toml) - Ethical hacking specialist\n- [**performance-engineer**](categories\u002F04-quality-security\u002Fperformance-engineer.toml) - Performance optimization expert\n- [**powershell-security-hardening**](categories\u002F04-quality-security\u002Fpowershell-security-hardening.toml) - PowerShell security hardening and compliance specialist\n- [**qa-expert**](categories\u002F04-quality-security\u002Fqa-expert.toml) - Test automation specialist\n- [**reviewer**](categories\u002F04-quality-security\u002Freviewer.toml) - PR-style review for correctness, security, and regressions\n- [**security-auditor**](categories\u002F04-quality-security\u002Fsecurity-auditor.toml) - Security vulnerability expert\n- [**test-automator**](categories\u002F04-quality-security\u002Ftest-automator.toml) - Test automation framework expert\n\n\u003C\u002Fdetails>\n\n\u003Cdetails>\n\u003Csummary>\u003Cb>05. Data & AI\u003C\u002Fb> — Data engineering, ML, and AI specialists (12 agents)\u003C\u002Fsummary>\n\n### [05. Data & AI](categories\u002F05-data-ai\u002F)\n\n- [**ai-engineer**](categories\u002F05-data-ai\u002Fai-engineer.toml) - AI system design and deployment expert\n- [**data-analyst**](categories\u002F05-data-ai\u002Fdata-analyst.toml) - Data insights and visualization specialist\n- [**data-engineer**](categories\u002F05-data-ai\u002Fdata-engineer.toml) - Data pipeline architect\n- [**data-scientist**](categories\u002F05-data-ai\u002Fdata-scientist.toml) - Analytics and insights expert\n- [**database-optimizer**](categories\u002F05-data-ai\u002Fdatabase-optimizer.toml) - Database performance specialist\n- [**llm-architect**](categories\u002F05-data-ai\u002Fllm-architect.toml) - Large language model architect\n- [**machine-learning-engineer**](categories\u002F05-data-ai\u002Fmachine-learning-engineer.toml) - Machine learning systems expert\n- [**ml-engineer**](categories\u002F05-data-ai\u002Fml-engineer.toml) - Machine learning specialist\n- [**mlops-engineer**](categories\u002F05-data-ai\u002Fmlops-engineer.toml) - MLOps and model deployment expert\n- [**nlp-engineer**](categories\u002F05-data-ai\u002Fnlp-engineer.toml) - Natural language processing expert\n- [**postgres-pro**](categories\u002F05-data-ai\u002Fpostgres-pro.toml) - PostgreSQL database expert\n- [**prompt-engineer**](categories\u002F05-data-ai\u002Fprompt-engineer.toml) - Prompt optimization specialist\n\n\u003C\u002Fdetails>\n\n\u003Cdetails>\n\u003Csummary>\u003Cb>06. Developer Experience\u003C\u002Fb> — Tooling and developer productivity experts (13 agents)\u003C\u002Fsummary>\n\n### [06. Developer Experience](categories\u002F06-developer-experience\u002F)\n\n- [**build-engineer**](categories\u002F06-developer-experience\u002Fbuild-engineer.toml) - Build system specialist\n- [**cli-developer**](categories\u002F06-developer-experience\u002Fcli-developer.toml) - Command-line tool creator\n- [**dependency-manager**](categories\u002F06-developer-experience\u002Fdependency-manager.toml) - Package and dependency specialist\n- [**documentation-engineer**](categories\u002F06-developer-experience\u002Fdocumentation-engineer.toml) - Technical documentation expert\n- [**dx-optimizer**](categories\u002F06-developer-experience\u002Fdx-optimizer.toml) - Developer experience optimization specialist\n- [**git-workflow-manager**](categories\u002F06-developer-experience\u002Fgit-workflow-manager.toml) - Git workflow and branching expert\n- [**legacy-modernizer**](categories\u002F06-developer-experience\u002Flegacy-modernizer.toml) - Legacy code modernization specialist\n- [**mcp-developer**](categories\u002F06-developer-experience\u002Fmcp-developer.toml) - Model Context Protocol specialist\n- [**powershell-module-architect**](categories\u002F06-developer-experience\u002Fpowershell-module-architect.toml) - PowerShell module and profile architecture specialist\n- [**powershell-ui-architect**](categories\u002F06-developer-experience\u002Fpowershell-ui-architect.toml) - PowerShell UI\u002FUX specialist for WinForms, WPF, Metro frameworks, and TUIs\n- [**refactoring-specialist**](categories\u002F06-developer-experience\u002Frefactoring-specialist.toml) - Code refactoring expert\n- [**slack-expert**](categories\u002F06-developer-experience\u002Fslack-expert.toml) - Slack platform and @slack\u002Fbolt specialist\n- [**tooling-engineer**](categories\u002F06-developer-experience\u002Ftooling-engineer.toml) - Developer tooling specialist\n\n\u003C\u002Fdetails>\n\n\u003Cdetails>\n\u003Csummary>\u003Cb>07. Specialized Domains\u003C\u002Fb> — Domain-specific technology experts (12 agents)\u003C\u002Fsummary>\n\n### [07. Specialized Domains](categories\u002F07-specialized-domains\u002F)\n\n- [**api-documenter**](categories\u002F07-specialized-domains\u002Fapi-documenter.toml) - API documentation specialist\n- [**blockchain-developer**](categories\u002F07-specialized-domains\u002Fblockchain-developer.toml) - Web3 and crypto specialist\n- [**embedded-systems**](categories\u002F07-specialized-domains\u002Fembedded-systems.toml) - Embedded and real-time systems expert\n- [**fintech-engineer**](categories\u002F07-specialized-domains\u002Ffintech-engineer.toml) - Financial technology specialist\n- [**game-developer**](categories\u002F07-specialized-domains\u002Fgame-developer.toml) - Game development expert\n- [**iot-engineer**](categories\u002F07-specialized-domains\u002Fiot-engineer.toml) - IoT systems developer\n- [**m365-admin**](categories\u002F07-specialized-domains\u002Fm365-admin.toml) - Microsoft 365, Exchange Online, Teams, and SharePoint administration specialist\n- [**mobile-app-developer**](categories\u002F07-specialized-domains\u002Fmobile-app-developer.toml) - Mobile application specialist\n- [**payment-integration**](categories\u002F07-specialized-domains\u002Fpayment-integration.toml) - Payment systems expert\n- [**quant-analyst**](categories\u002F07-specialized-domains\u002Fquant-analyst.toml) - Quantitative analysis specialist\n- [**risk-manager**](categories\u002F07-specialized-domains\u002Frisk-manager.toml) - Risk assessment and management expert\n- [**seo-specialist**](categories\u002F07-specialized-domains\u002Fseo-specialist.toml) - Search engine optimization expert\n\n\u003C\u002Fdetails>\n\n\u003Cdetails>\n\u003Csummary>\u003Cb>08. Business & Product\u003C\u002Fb> — Product management and business analysis (11 agents)\u003C\u002Fsummary>\n\n### [08. Business & Product](categories\u002F08-business-product\u002F)\n\n- [**business-analyst**](categories\u002F08-business-product\u002Fbusiness-analyst.toml) - Requirements specialist\n- [**content-marketer**](categories\u002F08-business-product\u002Fcontent-marketer.toml) - Content marketing specialist\n- [**customer-success-manager**](categories\u002F08-business-product\u002Fcustomer-success-manager.toml) - Customer success expert\n- [**legal-advisor**](categories\u002F08-business-product\u002Flegal-advisor.toml) - Legal and compliance specialist\n- [**product-manager**](categories\u002F08-business-product\u002Fproduct-manager.toml) - Product strategy expert\n- [**project-manager**](categories\u002F08-business-product\u002Fproject-manager.toml) - Project management specialist\n- [**sales-engineer**](categories\u002F08-business-product\u002Fsales-engineer.toml) - Technical sales expert\n- [**scrum-master**](categories\u002F08-business-product\u002Fscrum-master.toml) - Agile methodology expert\n- [**technical-writer**](categories\u002F08-business-product\u002Ftechnical-writer.toml) - Technical documentation specialist\n- [**ux-researcher**](categories\u002F08-business-product\u002Fux-researcher.toml) - User research expert\n- [**wordpress-master**](categories\u002F08-business-product\u002Fwordpress-master.toml) - WordPress development and optimization expert\n\n\u003C\u002Fdetails>\n\n\u003Cdetails>\n\u003Csummary>\u003Cb>09. Meta & Orchestration\u003C\u002Fb> — Agent coordination and meta-programming (12 agents)\u003C\u002Fsummary>\n\n### [09. Meta & Orchestration](categories\u002F09-meta-orchestration\u002F)\n\n- [**agent-installer**](categories\u002F09-meta-orchestration\u002Fagent-installer.toml) - Browse and install agents from this repository via GitHub\n- [**agent-organizer**](categories\u002F09-meta-orchestration\u002Fagent-organizer.toml) - Multi-agent coordinator\n- [**context-manager**](categories\u002F09-meta-orchestration\u002Fcontext-manager.toml) - Context optimization expert\n- [**error-coordinator**](categories\u002F09-meta-orchestration\u002Ferror-coordinator.toml) - Error handling and recovery specialist\n- [**it-ops-orchestrator**](categories\u002F09-meta-orchestration\u002Fit-ops-orchestrator.toml) - IT operations workflow orchestration specialist\n- [**knowledge-synthesizer**](categories\u002F09-meta-orchestration\u002Fknowledge-synthesizer.toml) - Knowledge aggregation expert\n- [**multi-agent-coordinator**](categories\u002F09-meta-orchestration\u002Fmulti-agent-coordinator.toml) - Advanced multi-agent orchestration\n- [**performance-monitor**](categories\u002F09-meta-orchestration\u002Fperformance-monitor.toml) - Agent performance optimization\n- [**pied-piper**](https:\u002F\u002Fgithub.com\u002Fsathish316\u002Fpied-piper\u002F) - Orchestrate Team of AI Subagents for repetitive SDLC workflows\n- [**task-distributor**](categories\u002F09-meta-orchestration\u002Ftask-distributor.toml) - Task allocation specialist\n- [**workflow-orchestrator**](categories\u002F09-meta-orchestration\u002Fworkflow-orchestrator.toml) - Complex workflow automation\n\n\u003C\u002Fdetails>\n\n\u003Cdetails>\n\u003Csummary>\u003Cb>10. Research & Analysis\u003C\u002Fb> — Research, search, and analysis specialists (7 agents)\u003C\u002Fsummary>\n\n### [10. Research & Analysis](categories\u002F10-research-analysis\u002F)\n\n- [**competitive-analyst**](categories\u002F10-research-analysis\u002Fcompetitive-analyst.toml) - Competitive intelligence specialist\n- [**data-researcher**](categories\u002F10-research-analysis\u002Fdata-researcher.toml) - Data discovery and analysis expert\n- [**docs-researcher**](categories\u002F10-research-analysis\u002Fdocs-researcher.toml) - Documentation-backed API and framework verification\n- [**market-researcher**](categories\u002F10-research-analysis\u002Fmarket-researcher.toml) - Market analysis and consumer insights\n- [**research-analyst**](categories\u002F10-research-analysis\u002Fresearch-analyst.toml) - Comprehensive research specialist\n- [**search-specialist**](categories\u002F10-research-analysis\u002Fsearch-specialist.toml) - Advanced information retrieval expert\n- [**trend-analyst**](categories\u002F10-research-analysis\u002Ftrend-analyst.toml) - Emerging trends and forecasting expert\n\n\u003C\u002Fdetails>\n\n## Understanding Subagents\n\nSubagents are specialized AI assistants that enhance Codex's capabilities by providing task-specific expertise. They act as dedicated helpers that Codex can call upon when encountering particular types of work.\n\n### What Makes Subagents Special?\n\n**Independent Context Windows**\nEvery subagent operates within its own isolated context space, preventing cross-contamination between different tasks and maintaining clarity in the primary conversation thread.\n\n**Domain-Specific Intelligence**\nSubagents come equipped with carefully crafted instructions tailored to their area of expertise, resulting in superior performance on specialized tasks.\n\n**Shared Across Projects**\nAfter creating a subagent, you can utilize it throughout various projects and distribute it among team members to ensure consistent development practices.\n\n**Explicit Delegation**\nCodex does not spawn subagents automatically. Use explicit delegation prompts to specify which agents to spawn, how to divide the work, and what shape the result should take.\n\n### Core Advantages\n\n- **Memory Efficiency**: Isolated contexts prevent the main conversation from becoming cluttered with task-specific details\n- **Enhanced Accuracy**: Specialized prompts and configurations lead to better results in specific domains\n- **Workflow Consistency**: Team-wide subagent sharing ensures uniform approaches to common tasks\n- **Codex-Native**: Uses `.toml` agent files aligned with official Codex subagent docs\n\n### Example Workflows\n\n**PR review workflow:**\n```text\nReview this branch with parallel subagents. Have reviewer look for correctness, security, and missing tests. Have docs_researcher verify the framework APIs this patch depends on. Wait for both and summarize the findings with file references.\n```\n\n**Bug investigation workflow:**\n```text\nInvestigate the broken settings flow. Have code_mapper trace the owning code paths, browser_debugger reproduce the bug in the browser, and frontend_developer propose the smallest fix after the failure is understood. Wait for the read-heavy agents first, then continue.\n```\n\n**Repo exploration and planning workflow:**\n```text\nUse search_specialist to locate the code related to payment retries, knowledge_synthesizer to summarize the current design, and refactoring_specialist to propose a minimal refactor plan. Return a concrete action list.\n```\n## Contributing\n\nWe welcome contributions! See [CONTRIBUTING.md](CONTRIBUTING.md) for guidelines.\n\n- Submit new subagents via PR\n- Improve existing definitions\n- Report issues and bugs\n\n\n## License\n\nMIT License - see [LICENSE](LICENSE)\n\nThis repository is a curated collection of subagent definitions contributed by both the maintainers and the community. All subagents are provided \"as is\" without warranty. We do not audit or guarantee the security or correctness of any subagent. Review before use, the maintainers accept no liability for any issues arising from their use.\n\nIf you find an issue with a listed subagent or want your contribution removed, please open an issue in this repository and we'll address it promptly.\n","\u003Ca href=\"https:\u002F\u002Fgithub.com\u002FVoltAgent\u002Fvoltagent\">\n    \u003Cimg width=\"1500\" height=\"500\" alt=\"codex\" src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FVoltAgent_awesome-codex-subagents_readme_5328e2808a49.png\" \u002F>\n\u003C\u002Fa>\n\n\u003Cbr \u002F>\n\u003Cbr \u002F>\n\n\u003Cdiv align=\"center\">\n    \u003Cstrong>涵盖10大类、超过136个Codex子代理的精彩合集。\u003C\u002Fstrong>\n    \u003Cbr \u002F>\n    \u003Cbr \u002F>\n\u003C\u002Fdiv>\n\n   \n\u003Cdiv align=\"center\">\n    \n[![Awesome](https:\u002F\u002Fawesome.re\u002Fbadge.svg)](https:\u002F\u002Fawesome.re)\n![子代理数量](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fsubagents-136-blue?style=classic)\n[![最近更新](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Flast-commit\u002FVoltAgent\u002Fawesome-codex-subagents?label=Last%20update&style=classic)](https:\u002F\u002Fgithub.com\u002FVoltAgent\u002Fawesome-codex-subagents)\n[![Discord](https:\u002F\u002Fimg.shields.io\u002Fdiscord\u002F1361559153780195478.svg?label=&logo=discord&logoColor=ffffff&color=7389D8&labelColor=6A7EC2)](https:\u002F\u002Fs.voltagent.dev\u002Fdiscord)\n\n\u003Cbr \u002F>\n\n\n\u003Cdiv align=\"center\">\n    \u003Cstrong>更多面向开发者的精彩合集\u003C\u002Fstrong>\n    \u003Cbr \u002F>\n    \u003Cbr \u002F>\n\u003C\u002Fdiv>\n\n[![代理技能](https:\u002F\u002Fimg.shields.io\u002Fstatic\u002Fv1?label=%E2%9A%A1%20Agent&message=Skills%2012k&color=black&style=classic)](https:\u002F\u002Fgithub.com\u002FVoltAgent\u002Fawesome-agent-skills)\n[![Claude代码子代理](https:\u002F\u002Fimg.shields.io\u002Fstatic\u002Fv1?label=Claude&message=Code%20Subagents%2014k&color=D97757&style=classic&logo=claude&logoColor=D97757)](https:\u002F\u002Fgithub.com\u002FVoltAgent\u002Fawesome-claude-code-subagents)\n[![OpenClaw技能](https:\u002F\u002Fimg.shields.io\u002Fstatic\u002Fv1?label=%F0%9F%A6%9E%20OpenClaw&message=Skills%2040k&color=f53e36&style=classic)](https:\u002F\u002Fgithub.com\u002FVoltAgent\u002Fawesome-openclaw-skills)\n[![AI代理论文](https:\u002F\u002Fimg.shields.io\u002Fstatic\u002Fv1?label=arxiv&message=Agent%20Papers%20328&color=b31b1b&style=classic&logo=arxiv)](https:\u002F\u002Fgithub.com\u002FVoltAgent\u002Fawesome-ai-agent-papers)\n\u003C\u002Fdiv>\n\n\n# 令人惊叹的Codex子代理\n\n本仓库是[Codex子代理](https:\u002F\u002Fdevelopers.openai.com\u002Fcodex\u002Fsubagents)的权威合集，这些子代理是专为特定开发任务设计的AI助手。它们专为Codex打造，并与官方文档保持一致。\n\n## 安装\n\n请严格按照文档说明使用Codex自定义代理目录：\n\n- `~\u002F.codex\u002Fagents\u002F` 用于全局代理（在所有项目中可用）\n- `.codex\u002Fagents\u002F` 用于项目专用代理（在该仓库中优先级更高）\n\n1. 克隆本仓库。\n2. 将您需要的`.toml`代理文件复制到上述任一目录中。\n3. 如有必要，重启或刷新您的Codex会话。\n4. 在提示中明确委派（Codex不会自动启动自定义子代理）。\n\n示例：\n```bash\nmkdir -p ~\u002F.codex\u002Fagents\ncp categories\u002F01-core-development\u002Fbackend-developer.toml ~\u002F.codex\u002Fagents\u002F\n```\n\n```bash\nmkdir -p .codex\u002Fagents\ncp categories\u002F04-quality-security\u002Freviewer.toml .codex\u002Fagents\u002F\n```\n\n如果您在Codex中使用代理配置，请将其保留在`.codex\u002Fconfig.toml`的`[agents]`部分，具体说明见官方文档。\n\n\n### 子代理存储位置\n\n| 类型 | 路径 | 可用性 | 优先级 |\n|------|------|--------------|------------|\n| 项目子代理 | `.codex\u002Fagents\u002F` | 仅当前项目 | 更高 |\n| 全局子代理 | `~\u002F.codex\u002Fagents\u002F` | 所有项目 | 更低 |\n\n注意：当出现命名冲突时，项目专用子代理将覆盖全局子代理。\n\n\n## 子代理结构\n\n每个子代理都采用Codex原生的`.toml`格式：\n\n```toml\nname = \"subagent-name\"\ndescription = \"何时应调用此代理\"\nmodel = \"gpt-5.3-codex-spark\"\nmodel_reasoning_effort = \"medium\"\nsandbox_mode = \"read-only\"\n\n[instructions]\ntext = \"\"\"\n你是一名[角色描述和专业领域]...\n\n[代理特有的检查清单、模式和指南]...\n\"\"\"\n```\n\n### 智能模型路由\n\n每个子代理都包含一个`model`字段，可自动将其路由到合适的模型——在质量和成本之间取得平衡：\n\n| 模型 | 使用场景 | 示例 |\n|-------|----------------|----------|\n| `gpt-5.4` | 深度推理——架构评审、安全审计、金融逻辑 | `security-auditor`、`architect-reviewer`、`fintech-engineer` |\n| `gpt-5.3-codex-spark` | 快速扫描、综合及较轻的研究任务 | `search-specialist`、`docs-researcher`、`agent-installer` |\n\n### 沙盒模式理念\n\n每个子代理的`sandbox_mode`字段控制其对文件系统的访问权限：\n- **只读代理**（评审员、审计员）：`sandbox_mode = \"read-only\"`——仅分析而不修改\n- **工作区写入代理**（开发者、工程师）：`sandbox_mode = \"workspace-write\"`——创建和修改文件\n\n\n## 分类\n\n### [01. 核心开发](categories\u002F01-core-development\u002F)\n\n适用于日常编码任务的核心开发子代理。\n\n- [**api-designer**](categories\u002F01-core-development\u002Fapi-designer.toml)——REST和GraphQL API架构师\n- [**backend-developer**](categories\u002F01-core-development\u002Fbackend-developer.toml)——服务器端专家，负责构建可扩展的API\n- [**code-mapper**](categories\u002F01-core-development\u002Fcode-mapper.toml)——代码路径映射及所有权边界分析\n- [**electron-pro**](categories\u002F01-core-development\u002Felectron-pro.toml)——桌面应用专家\n- [**frontend-developer**](categories\u002F01-core-development\u002Ffrontend-developer.toml)——React、Vue和Angular的UI\u002FUX专家\n- [**fullstack-developer**](categories\u002F01-core-development\u002Ffullstack-developer.toml)——全栈功能开发\n- [**graphql-architect**](categories\u002F01-core-development\u002Fgraphql-architect.toml)——GraphQL模式和联邦专家\n- [**microservices-architect**](categories\u002F01-core-development\u002Fmicroservices-architect.toml)——分布式系统设计师\n- [**mobile-developer**](categories\u002F01-core-development\u002Fmobile-developer.toml)——跨平台移动应用专家\n- [**ui-designer**](categories\u002F01-core-development\u002Fui-designer.toml)——视觉设计和交互专家\n- [**ui-fixer**](categories\u002F01-core-development\u002Fui-fixer.toml)——针对已复现UI问题的最小安全修复\n- [**websocket-engineer**](categories\u002F01-core-development\u002Fwebsocket-engineer.toml)——实时通信专家\n\n### [02. 语言专家](categories\u002F02-language-specialists\u002F)\n\n具备深厚框架知识的语言特定专家。\n- [**angular-architect**](categories\u002F02-language-specialists\u002Fangular-architect.toml) - Angular 15+ 企业级模式专家\n- [**cpp-pro**](categories\u002F02-language-specialists\u002Fcpp-pro.toml) - C++ 性能专家\n- [**csharp-developer**](categories\u002F02-language-specialists\u002Fcsharp-developer.toml) - .NET 生态系统专家\n- [**django-developer**](categories\u002F02-language-specialists\u002Fdjango-developer.toml) - Django 4+ Web 开发专家\n- [**dotnet-core-expert**](categories\u002F02-language-specialists\u002Fdotnet-core-expert.toml) - .NET 8 跨平台专家\n- [**dotnet-framework-4.8-expert**](categories\u002F02-language-specialists\u002Fdotnet-framework-4.8-expert.toml) - .NET Framework 遗留企业级专家\n- [**elixir-expert**](categories\u002F02-language-specialists\u002Felixir-expert.toml) - Elixir 和 OTP 容错系统专家\n- [**erlang-expert**](categories\u002F02-language-specialists\u002Ferlang-expert.toml) - Erlang\u002FOTP 和 rebar3 工程专家\n- [**flutter-expert**](categories\u002F02-language-specialists\u002Fflutter-expert.toml) - Flutter 3+ 跨平台移动开发专家\n- [**golang-pro**](categories\u002F02-language-specialists\u002Fgolang-pro.toml) - Go 并发编程专家\n- [**java-architect**](categories\u002F02-language-specialists\u002Fjava-architect.toml) - 企业级 Java 专家\n- [**javascript-pro**](categories\u002F02-language-specialists\u002Fjavascript-pro.toml) - JavaScript 开发专家\n- [**kotlin-specialist**](categories\u002F02-language-specialists\u002Fkotlin-specialist.toml) - 现代 JVM 语言专家\n- [**laravel-specialist**](categories\u002F02-language-specialists\u002Flaravel-specialist.toml) - Laravel 10+ PHP 框架专家\n- [**nextjs-developer**](categories\u002F02-language-specialists\u002Fnextjs-developer.toml) - Next.js 14+ 全栈专家\n- [**php-pro**](categories\u002F02-language-specialists\u002Fphp-pro.toml) - PHP Web 开发专家\n- [**powershell-5.1-expert**](categories\u002F02-language-specialists\u002Fpowershell-5.1-expert.toml) - Windows PowerShell 5.1 和完整 .NET Framework 自动化专家\n- [**powershell-7-expert**](categories\u002F02-language-specialists\u002Fpowershell-7-expert.toml) - 跨平台 PowerShell 7+ 自动化及现代 .NET 专家\n- [**python-pro**](categories\u002F02-language-specialists\u002Fpython-pro.toml) - Python 生态系统大师\n- [**rails-expert**](categories\u002F02-language-specialists\u002Frails-expert.toml) - Rails 8.1 快速开发专家\n- [**react-specialist**](categories\u002F02-language-specialists\u002Freact-specialist.toml) - React 18+ 现代模式专家\n- [**rust-engineer**](categories\u002F02-language-specialists\u002Frust-engineer.toml) - 系统编程专家\n- [**spring-boot-engineer**](categories\u002F02-language-specialists\u002Fspring-boot-engineer.toml) - Spring Boot 3+ 微服务专家\n- [**sql-pro**](categories\u002F02-language-specialists\u002Fsql-pro.toml) - 数据库查询专家\n- [**swift-expert**](categories\u002F02-language-specialists\u002Fswift-expert.toml) - iOS 和 macOS 专家\n- [**typescript-pro**](categories\u002F02-language-specialists\u002Ftypescript-pro.toml) - TypeScript 专家\n- [**vue-expert**](categories\u002F02-language-specialists\u002Fvue-expert.toml) - Vue 3 Composition API 专家\n\n\n### [03. 基础设施](categories\u002F03-infrastructure\u002F)\n\nDevOps、云和部署专家。\n\n- [**azure-infra-engineer**](categories\u002F03-infrastructure\u002Fazure-infra-engineer.toml) - Azure 基础设施和 Az PowerShell 自动化专家\n- [**cloud-architect**](categories\u002F03-infrastructure\u002Fcloud-architect.toml) - AWS\u002FGCP\u002FAzure 专家\n- [**database-administrator**](categories\u002F03-infrastructure\u002Fdatabase-administrator.toml) - 数据库管理专家\n- [**deployment-engineer**](categories\u002F03-infrastructure\u002Fdeployment-engineer.toml) - 部署自动化专家\n- [**devops-engineer**](categories\u002F03-infrastructure\u002Fdevops-engineer.toml) - CI\u002FCD 和自动化专家\n- [**devops-incident-responder**](categories\u002F03-infrastructure\u002Fdevops-incident-responder.toml) - DevOps 事件管理\n- [**docker-expert**](categories\u002F03-infrastructure\u002Fdocker-expert.toml) - Docker 容器化和优化专家\n- [**incident-responder**](categories\u002F03-infrastructure\u002Fincident-responder.toml) - 系统事件响应专家\n- [**kubernetes-specialist**](categories\u002F03-infrastructure\u002Fkubernetes-specialist.toml) - 容器编排大师\n- [**network-engineer**](categories\u002F03-infrastructure\u002Fnetwork-engineer.toml) - 网络基础设施专家\n- [**platform-engineer**](categories\u002F03-infrastructure\u002Fplatform-engineer.toml) - 平台架构专家\n- [**security-engineer**](categories\u002F03-infrastructure\u002Fsecurity-engineer.toml) - 基础设施安全专家\n- [**sre-engineer**](categories\u002F03-infrastructure\u002Fsre-engineer.toml) - 站点可靠性工程专家\n- [**terraform-engineer**](categories\u002F03-infrastructure\u002Fterraform-engineer.toml) - Infrastructure as Code 专家\n- [**terragrunt-expert**](categories\u002F03-infrastructure\u002Fterragrunt-expert.toml) - Terragrunt 编排和 DRY IaC 专家\n- [**windows-infra-admin**](categories\u002F03-infrastructure\u002Fwindows-infra-admin.toml) - Active Directory、DNS、DHCP 和 GPO 自动化专家\n\n\u003Cdetails>\n\u003Csummary>\u003Cb>04. 质量与安全\u003C\u002Fb> — 测试、安全和代码质量专家（16 个代理）\u003C\u002Fsummary>\n\n### [04. 质量与安全](categories\u002F04-quality-security\u002F)\n\n- [**accessibility-tester**](categories\u002F04-quality-security\u002Faccessibility-tester.toml) - 无障碍合规专家\n- [**ad-security-reviewer**](categories\u002F04-quality-security\u002Fad-security-reviewer.toml) - Active Directory 安全与 GPO 审计专家\n- [**architect-reviewer**](categories\u002F04-quality-security\u002Farchitect-reviewer.toml) - 架构评审专家\n- [**browser-debugger**](categories\u002F04-quality-security\u002Fbrowser-debugger.toml) - 基于浏览器的复现与客户端调试\n- [**chaos-engineer**](categories\u002F04-quality-security\u002Fchaos-engineer.toml) - 系统韧性测试专家\n- [**code-reviewer**](categories\u002F04-quality-security\u002Fcode-reviewer.toml) - 代码质量守护者\n- [**compliance-auditor**](categories\u002F04-quality-security\u002Fcompliance-auditor.toml) - 法规合规专家\n- [**debugger**](categories\u002F04-quality-security\u002Fdebugger.toml) - 高级调试专家\n- [**error-detective**](categories\u002F04-quality-security\u002Ferror-detective.toml) - 错误分析与解决专家\n- [**penetration-tester**](categories\u002F04-quality-security\u002Fpenetration-tester.toml) - 道德黑客专家\n- [**performance-engineer**](categories\u002F04-quality-security\u002Fperformance-engineer.toml) - 性能优化专家\n- [**powershell-security-hardening**](categories\u002F04-quality-security\u002Fpowershell-security-hardening.toml) - PowerShell 安全加固与合规专家\n- [**qa-expert**](categories\u002F04-quality-security\u002Fqa-expert.toml) - 测试自动化专家\n- [**reviewer**](categories\u002F04-quality-security\u002Freviewer.toml) - PR 风格的正确性、安全性及回归测试评审\n- [**security-auditor**](categories\u002F04-quality-security\u002Fsecurity-auditor.toml) - 安全漏洞专家\n- [**test-automator**](categories\u002F04-quality-security\u002Ftest-automator.toml) - 测试自动化框架专家\n\n\u003C\u002Fdetails>\n\n\u003Cdetails>\n\u003Csummary>\u003Cb>05. 数据与人工智能\u003C\u002Fb> — 数据工程、机器学习和人工智能专家（12个代理）\u003C\u002Fsummary>\n\n### [05. 数据与人工智能](categories\u002F05-data-ai\u002F)\n\n- [**ai-engineer**](categories\u002F05-data-ai\u002Fai-engineer.toml) - AI 系统设计与部署专家\n- [**data-analyst**](categories\u002F05-data-ai\u002Fdata-analyst.toml) - 数据洞察与可视化专家\n- [**data-engineer**](categories\u002F05-data-ai\u002Fdata-engineer.toml) - 数据管道架构师\n- [**data-scientist**](categories\u002F05-data-ai\u002Fdata-scientist.toml) - 分析与洞察专家\n- [**database-optimizer**](categories\u002F05-data-ai\u002Fdatabase-optimizer.toml) - 数据库性能专家\n- [**llm-architect**](categories\u002F05-data-ai\u002Fllm-architect.toml) - 大型语言模型架构师\n- [**machine-learning-engineer**](categories\u002F05-data-ai\u002Fmachine-learning-engineer.toml) - 机器学习系统专家\n- [**ml-engineer**](categories\u002F05-data-ai\u002Fml-engineer.toml) - 机器学习专家\n- [**mlops-engineer**](categories\u002F05-data-ai\u002Fmlops-engineer.toml) - MLOps 和模型部署专家\n- [**nlp-engineer**](categories\u002F05-data-ai\u002Fnlp-engineer.toml) - 自然语言处理专家\n- [**postgres-pro**](categories\u002F05-data-ai\u002Fpostgres-pro.toml) - PostgreSQL 数据库专家\n- [**prompt-engineer**](categories\u002F05-data-ai\u002Fprompt-engineer.toml) - 提示词优化专家\n\n\u003C\u002Fdetails>\n\n\u003Cdetails>\n\u003Csummary>\u003Cb>06. 开发者体验\u003C\u002Fb> — 工具链与开发者生产力专家（13个代理）\u003C\u002Fsummary>\n\n### [06. 开发者体验](categories\u002F06-developer-experience\u002F)\n\n- [**build-engineer**](categories\u002F06-developer-experience\u002Fbuild-engineer.toml) - 构建系统专家\n- [**cli-developer**](categories\u002F06-developer-experience\u002Fcli-developer.toml) - 命令行工具开发者\n- [**dependency-manager**](categories\u002F06-developer-experience\u002Fdependency-manager.toml) - 包与依赖管理专家\n- [**documentation-engineer**](categories\u002F06-developer-experience\u002Fdocumentation-engineer.toml) - 技术文档专家\n- [**dx-optimizer**](categories\u002F06-developer-experience\u002Fdx-optimizer.toml) - 开发者体验优化专家\n- [**git-workflow-manager**](categories\u002F06-developer-experience\u002Fgit-workflow-manager.toml) - Git 工作流与分支管理专家\n- [**legacy-modernizer**](categories\u002F06-developer-experience\u002Flegacy-modernizer.toml) - 旧代码现代化专家\n- [**mcp-developer**](categories\u002F06-developer-experience\u002Fmcp-developer.toml) - 模型上下文协议专家\n- [**powershell-module-architect**](categories\u002F06-developer-experience\u002Fpowershell-module-architect.toml) - PowerShell 模块与配置文件架构专家\n- [**powershell-ui-architect**](categories\u002F06-developer-experience\u002Fpowershell-ui-architect.toml) - PowerShell UI\u002FUX 专家，适用于 WinForms、WPF、Metro 框架及 TUI\n- [**refactoring-specialist**](categories\u002F06-developer-experience\u002Frefactoring-specialist.toml) - 代码重构专家\n- [**slack-expert**](categories\u002F06-developer-experience\u002Fslack-expert.toml) - Slack 平台与 @slack\u002Fbolt 专家\n- [**tooling-engineer**](categories\u002F06-developer-experience\u002Ftooling-engineer.toml) - 开发者工具链专家\n\n\u003C\u002Fdetails>\n\n\u003Cdetails>\n\u003Csummary>\u003Cb>07. 专业领域\u003C\u002Fb> — 领域特定技术专家（12个代理）\u003C\u002Fsummary>\n\n### [07. 专业领域](categories\u002F07-specialized-domains\u002F)\n\n- [**api-documenter**](categories\u002F07-specialized-domains\u002Fapi-documenter.toml) - API 文档专家\n- [**blockchain-developer**](categories\u002F07-specialized-domains\u002Fblockchain-developer.toml) - Web3 和加密货币专家\n- [**embedded-systems**](categories\u002F07-specialized-domains\u002Fembedded-systems.toml) - 嵌入式与实时系统专家\n- [**fintech-engineer**](categories\u002F07-specialized-domains\u002Ffintech-engineer.toml) - 金融科技专家\n- [**game-developer**](categories\u002F07-specialized-domains\u002Fgame-developer.toml) - 游戏开发专家\n- [**iot-engineer**](categories\u002F07-specialized-domains\u002Fiot-engineer.toml) - IoT 系统开发者\n- [**m365-admin**](categories\u002F07-specialized-domains\u002Fm365-admin.toml) - Microsoft 365、Exchange Online、Teams 和 SharePoint 管理专家\n- [**mobile-app-developer**](categories\u002F07-specialized-domains\u002Fmobile-app-developer.toml) - 移动应用专家\n- [**payment-integration**](categories\u002F07-specialized-domains\u002Fpayment-integration.toml) - 支付系统专家\n- [**quant-analyst**](categories\u002F07-specialized-domains\u002Fquant-analyst.toml) - 定量分析专家\n- [**risk-manager**](categories\u002F07-specialized-domains\u002Frisk-manager.toml) - 风险评估与管理专家\n- [**seo-specialist**](categories\u002F07-specialized-domains\u002Fseo-specialist.toml) - 搜索引擎优化专家\n\n\u003C\u002Fdetails>\n\n\u003Cdetails>\n\u003Csummary>\u003Cb>08. 商业与产品\u003C\u002Fb> — 产品管理和商业分析（11个代理）\u003C\u002Fsummary>\n\n### [08. 业务与产品](categories\u002F08-business-product\u002F)\n\n- [**business-analyst**](categories\u002F08-business-product\u002Fbusiness-analyst.toml) - 需求分析师\n- [**content-marketer**](categories\u002F08-business-product\u002Fcontent-marketer.toml) - 内容营销专家\n- [**customer-success-manager**](categories\u002F08-business-product\u002Fcustomer-success-manager.toml) - 客户成功专家\n- [**legal-advisor**](categories\u002F08-business-product\u002Flegal-advisor.toml) - 法务与合规专家\n- [**product-manager**](categories\u002F08-business-product\u002Fproduct-manager.toml) - 产品战略专家\n- [**project-manager**](categories\u002F08-business-product\u002Fproject-manager.toml) - 项目管理专家\n- [**sales-engineer**](categories\u002F08-business-product\u002Fsales-engineer.toml) - 技术销售专家\n- [**scrum-master**](categories\u002F08-business-product\u002Fscrum-master.toml) - 敏捷方法论专家\n- [**technical-writer**](categories\u002F08-business-product\u002Ftechnical-writer.toml) - 技术文档专家\n- [**ux-researcher**](categories\u002F08-business-product\u002Fux-researcher.toml) - 用户研究专家\n- [**wordpress-master**](categories\u002F08-business-product\u002Fwordpress-master.toml) - WordPress开发与优化专家\n\n\u003C\u002Fdetails>\n\n\u003Cdetails>\n\u003Csummary>\u003Cb>09. 元与编排\u003C\u002Fb> — 代理协调与元编程（12个代理）\u003C\u002Fsummary>\n\n### [09. 元与编排](categories\u002F09-meta-orchestration\u002F)\n\n- [**agent-installer**](categories\u002F09-meta-orchestration\u002Fagent-installer.toml) - 通过GitHub从本仓库浏览并安装代理\n- [**agent-organizer**](categories\u002F09-meta-orchestration\u002Fagent-organizer.toml) - 多代理协调员\n- [**context-manager**](categories\u002F09-meta-orchestration\u002Fcontext-manager.toml) - 上下文优化专家\n- [**error-coordinator**](categories\u002F09-meta-orchestration\u002Ferror-coordinator.toml) - 错误处理与恢复专家\n- [**it-ops-orchestrator**](categories\u002F09-meta-orchestration\u002Fit-ops-orchestrator.toml) - IT运维流程编排专家\n- [**knowledge-synthesizer**](categories\u002F09-meta-orchestration\u002Fknowledge-synthesizer.toml) - 知识聚合专家\n- [**multi-agent-coordinator**](categories\u002F09-meta-orchestration\u002Fmulti-agent-coordinator.toml) - 高级多代理编排\n- [**performance-monitor**](categories\u002F09-meta-orchestration\u002Fperformance-monitor.toml) - 代理性能优化\n- [**pied-piper**](https:\u002F\u002Fgithub.com\u002Fsathish316\u002Fpied-piper\u002F) - 为重复性SDLC工作流编排AI子代理团队\n- [**task-distributor**](categories\u002F09-meta-orchestration\u002Ftask-distributor.toml) - 任务分配专家\n- [**workflow-orchestrator**](categories\u002F09-meta-orchestration\u002Fworkflow-orchestrator.toml) - 复杂工作流自动化\n\n\u003C\u002Fdetails>\n\n\u003Cdetails>\n\u003Csummary>\u003Cb>10. 研究与分析\u003C\u002Fb> — 研究、搜索与分析专家（7个代理）\u003C\u002Fsummary>\n\n### [10. 研究与分析](categories\u002F10-research-analysis\u002F)\n\n- [**competitive-analyst**](categories\u002F10-research-analysis\u002Fcompetitive-analyst.toml) - 竞争情报专家\n- [**data-researcher**](categories\u002F10-research-analysis\u002Fdata-researcher.toml) - 数据发现与分析专家\n- [**docs-researcher**](categories\u002F10-research-analysis\u002Fdocs-researcher.toml) - 基于文档的API和框架验证\n- [**market-researcher**](categories\u002F10-research-analysis\u002Fmarket-researcher.toml) - 市场分析与消费者洞察\n- [**research-analyst**](categories\u002F10-research-analysis\u002Fresearch-analyst.toml) - 综合研究专家\n- [**search-specialist**](categories\u002F10-research-analysis\u002Fsearch-specialist.toml) - 高级信息检索专家\n- [**trend-analyst**](categories\u002F10-research-analysis\u002Ftrend-analyst.toml) - 新兴趋势与预测专家\n\n\u003C\u002Fdetails>\n\n## 子代理的理解\n\n子代理是专门化的AI助手，通过提供特定任务的专业知识来增强Codex的能力。它们充当Codex在遇到特定类型工作时可以调用的专用助手。\n\n### 子代理有何特别之处？\n\n**独立的上下文窗口**\n每个子代理都在自己的隔离上下文空间内运行，防止不同任务之间的交叉污染，并保持主对话线程的清晰度。\n\n**领域特定智能**\n子代理配备了针对其专业领域的精心设计指令，从而在专业任务上表现出色。\n\n**跨项目共享**\n创建子代理后，您可以在各个项目中使用它，并将其分发给团队成员，以确保一致的开发实践。\n\n**明确委托**\nCodex不会自动生成子代理。请使用明确的委托提示来指定要生成哪些代理、如何分工以及结果应呈现的形式。\n\n### 核心优势\n\n- **内存效率**：隔离的上下文可防止主对话因任务特定细节而变得混乱。\n- **更高的准确性**：专门化的提示和配置使子代理在特定领域取得更好的效果。\n- **工作流一致性**：团队范围内共享子代理可确保对常见任务采取统一的方法。\n- **原生支持Codex**：使用与官方Codex子代理文档一致的`.toml`代理文件。\n\n### 示例工作流\n\n**PR评审工作流：**\n```text\n使用并行子代理评审此分支。让reviewer检查代码正确性、安全性及是否缺少测试。让docs_researcher验证此补丁所依赖的框架API。等待两者完成，并附带文件引用总结发现。\n```\n\n**Bug调查工作流：**\n```text\n调查设置流程中的故障。让code_mapper追踪相关代码路径，browser_debugger在浏览器中复现该Bug，frontend_developer在理解故障原因后提出最小修复方案。先等待读取密集型代理完成，再继续。\n```\n\n**代码库探索与规划工作流：**\n```text\n使用search_specialist查找与支付重试相关的代码，knowledge_synthesizer总结当前设计，refactoring_specialist提出最小重构计划。返回具体的行动清单。\n```\n## 贡献\n\n我们欢迎您的贡献！请参阅[CONTRIBUTING.md](CONTRIBUTING.md)了解指南。\n\n- 通过PR提交新的子代理\n- 改进现有定义\n- 报告问题和漏洞\n\n\n## 许可证\n\nMIT许可证 - 请参阅[LICENSE](LICENSE)\n\n本仓库是由维护者和社区共同贡献的子代理定义精选集合。所有子代理均按“现状”提供，不提供任何保证。我们不对任何子代理的安全性或正确性进行审计或担保。使用前请仔细审查，维护者对因其使用而产生的任何问题概不负责。\n\n如果您发现已列出的子代理存在问题，或希望移除您的贡献，请在此仓库中提交问题，我们将尽快处理。","# awesome-codex-subagents 快速上手指南\n\n本指南帮助中国开发者快速集成 136+ 个专为 OpenAI Codex 设计的子代理（Subagents），涵盖核心开发、语言专家、基础设施及安全审计等十大类别。\n\n## 环境准备\n\n*   **系统要求**：支持 macOS、Linux 或 Windows (WSL)。\n*   **前置依赖**：\n    *   已安装并配置好 **OpenAI Codex** 命令行工具。\n    *   具备基础的终端操作能力。\n    *   确保已登录有效的 OpenAI 账号并具有 Codex 访问权限。\n\n## 安装步骤\n\n### 1. 克隆仓库\n首先将包含所有子代理配置的仓库克隆到本地：\n\n```bash\ngit clone https:\u002F\u002Fgithub.com\u002FVoltAgent\u002Fawesome-codex-subagents.git\ncd awesome-codex-subagents\n```\n\n> **提示**：如果国内访问 GitHub 较慢，可使用镜像源加速：\n> `git clone https:\u002F\u002Fghp.ci\u002Fhttps:\u002F\u002Fgithub.com\u002FVoltAgent\u002Fawesome-codex-subagents.git`\n\n### 2. 选择安装范围\nCodex 支持两种代理存储位置，请根据需求选择其一：\n\n*   **全局安装**（推荐）：所有项目均可使用。\n    ```bash\n    mkdir -p ~\u002F.codex\u002Fagents\n    # 示例：复制后端开发代理\n    cp categories\u002F01-core-development\u002Fbackend-developer.toml ~\u002F.codex\u002Fagents\u002F\n    ```\n\n*   **项目级安装**：仅当前项目可用（优先级更高）。\n    ```bash\n    mkdir -p .codex\u002Fagents\n    # 示例：复制代码审查代理\n    cp categories\u002F04-quality-security\u002Freviewer.toml .codex\u002Fagents\u002F\n    ```\n\n### 3. 生效配置\n复制完成后，重启终端或刷新当前的 Codex 会话即可生效。\n\n## 基本使用\n\nCodex **不会自动调用**自定义子代理，你需要在提示词（Prompt）中明确指定要调用的代理名称。\n\n### 使用示例\n\n假设你已安装了 `backend-developer` 代理，在终端输入 Codex 命令时，直接指明角色：\n\n```bash\ncodex \"请调用 backend-developer 代理，帮我设计一个高并发的用户认证 API，使用 Go 语言和 Gin 框架。\"\n```\n\n或者在项目目录中直接使用：\n\n```bash\ncodex \"使用 api-designer 代理为现有的订单服务生成 GraphQL Schema。\"\n```\n\n### 核心机制说明\n\n*   **智能模型路由**：无需手动指定模型。代理配置文件中的 `model` 字段会自动根据任务复杂度路由到合适的模型（如 `gpt-5.4` 用于深度架构分析，`gpt-5.3-codex-spark` 用于快速文档检索）。\n*   **沙箱安全控制**：\n    *   **只读模式** (`read-only`)：适用于审计、审查类代理（如 `security-auditor`），它们只能分析代码而不能修改文件。\n    *   **写入模式** (`workspace-write`)：适用于开发类代理（如 `frontend-developer`），允许创建和修改项目文件。\n\n通过明确调用特定的子代理名称，你可以获得比通用对话更专业、更符合特定技术栈的代码生成与建议。","某全栈开发团队正在紧急重构一个遗留的电商微服务系统，需要在两天内完成代码审查、安全漏洞修复及核心支付逻辑的重写。\n\n### 没有 awesome-codex-subagents 时\n- **角色切换成本高**：开发者需在同一个通用 AI 对话中反复提示“现在你是安全专家”或“现在你是后端架构师”，导致上下文混乱，AI 经常混淆任务目标。\n- **审查深度不足**：通用模型倾向于给出泛泛的代码建议，难以像专业审计员那样深入检查竞态条件或特定的金融逻辑漏洞。\n- **配置重复劳动**：每次开启新项目或新会话，都要重新编写复杂的 Prompt 来定义代码规范和检查清单，浪费大量启动时间。\n- **模型资源浪费**：无论任务难易都使用同一高算力模型，导致简单的格式调整也消耗高昂的 Token 成本，而复杂架构设计却可能因算力分配不足而思考不深。\n\n### 使用 awesome-codex-subagents 后\n- **即插即用专家团**：直接调用 `reviewer.toml` 和 `fintech-engineer.toml` 等预置子代理，AI 瞬间进入特定专家角色，无需任何额外的角色设定提示。\n- **深度专项攻坚**：`security-auditor` 自动路由至强推理模型（gpt-5.4），精准定位支付模块中的并发安全隐患；`backend-developer` 则高效处理常规业务逻辑重写。\n- **标准化作业流程**：每个子代理内置了详细的检查清单和执行规范，确保不同成员调用的 AI 输出风格一致、质量可控，消除人为提示词差异。\n- **智能成本优化**：利用子代理内置的智能路由机制，简单任务自动匹配轻量模型，复杂任务自动升级算力，在保障质量的同时显著降低运行成本。\n\nawesome-codex-subagents 将通用的 AI 助手转化为了一支分工明确、配置最优的虚拟专家团队，让开发者从“提示词工程师”回归到真正的“系统架构师”。","https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FVoltAgent_awesome-codex-subagents_5328e280.png","VoltAgent","https:\u002F\u002Foss.gittoolsai.com\u002Favatars\u002FVoltAgent_e51f56aa.png","AI Agent Engineering Platform built on an Open Source TypeScript AI Agent Framework",null,"info@voltagent.dev","voltagent_dev","https:\u002F\u002Fvoltagent.dev","https:\u002F\u002Fgithub.com\u002FVoltAgent",3829,437,"2026-04-12T02:17:23","MIT",2,"Linux, macOS, Windows","未说明",{"notes":31,"python":29,"dependencies":32},"本工具并非独立的 AI 模型或需要本地计算资源的软件，而是 OpenAI Codex CLI 的自定义子代理（Subagents）配置文件集合。运行前提是用户已安装并配置好 OpenAI Codex 命令行工具。安装方式是将 .toml 配置文件复制到指定目录（~\u002F.codex\u002Fagents\u002F 或项目下的 .codex\u002Fagents\u002F），无需安装 Python 依赖包、GPU 或特定内存。实际运行时的资源消耗取决于调用的云端模型（如 gpt-5.4 等）。",[33],"Codex CLI",[35,36],"Agent","语言模型",[38,39,40,41,42,43,44],"awesome-list","chatgpt","codex","codex-skills","codex-subagents","subagents","ai-agents","ready","2026-03-27T02:49:30.150509","2026-04-12T16:31:41.376975",[],[],[51,63,71,79,87,96],{"id":52,"name":53,"github_repo":54,"description_zh":55,"stars":56,"difficulty_score":57,"last_commit_at":58,"category_tags":59,"status":45},4358,"openclaw","openclaw\u002Fopenclaw","OpenClaw 是一款专为个人打造的本地化 AI 助手，旨在让你在自己的设备上拥有完全可控的智能伙伴。它打破了传统 AI 助手局限于特定网页或应用的束缚，能够直接接入你日常使用的各类通讯渠道，包括微信、WhatsApp、Telegram、Discord、iMessage 等数十种平台。无论你在哪个聊天软件中发送消息，OpenClaw 都能即时响应，甚至支持在 macOS、iOS 和 Android 设备上进行语音交互，并提供实时的画布渲染功能供你操控。\n\n这款工具主要解决了用户对数据隐私、响应速度以及“始终在线”体验的需求。通过将 AI 部署在本地，用户无需依赖云端服务即可享受快速、私密的智能辅助，真正实现了“你的数据，你做主”。其独特的技术亮点在于强大的网关架构，将控制平面与核心助手分离，确保跨平台通信的流畅性与扩展性。\n\nOpenClaw 非常适合希望构建个性化工作流的技术爱好者、开发者，以及注重隐私保护且不愿被单一生态绑定的普通用户。只要具备基础的终端操作能力（支持 macOS、Linux 及 Windows WSL2），即可通过简单的命令行引导完成部署。如果你渴望拥有一个懂你",349277,3,"2026-04-06T06:32:30",[35,60,61,62],"开发框架","图像","数据工具",{"id":64,"name":65,"github_repo":66,"description_zh":67,"stars":68,"difficulty_score":57,"last_commit_at":69,"category_tags":70,"status":45},3808,"stable-diffusion-webui","AUTOMATIC1111\u002Fstable-diffusion-webui","stable-diffusion-webui 是一个基于 Gradio 构建的网页版操作界面，旨在让用户能够轻松地在本地运行和使用强大的 Stable Diffusion 图像生成模型。它解决了原始模型依赖命令行、操作门槛高且功能分散的痛点，将复杂的 AI 绘图流程整合进一个直观易用的图形化平台。\n\n无论是希望快速上手的普通创作者、需要精细控制画面细节的设计师，还是想要深入探索模型潜力的开发者与研究人员，都能从中获益。其核心亮点在于极高的功能丰富度：不仅支持文生图、图生图、局部重绘（Inpainting）和外绘（Outpainting）等基础模式，还独创了注意力机制调整、提示词矩阵、负向提示词以及“高清修复”等高级功能。此外，它内置了 GFPGAN 和 CodeFormer 等人脸修复工具，支持多种神经网络放大算法，并允许用户通过插件系统无限扩展能力。即使是显存有限的设备，stable-diffusion-webui 也提供了相应的优化选项，让高质量的 AI 艺术创作变得触手可及。",162132,"2026-04-05T11:01:52",[60,61,35],{"id":72,"name":73,"github_repo":74,"description_zh":75,"stars":76,"difficulty_score":27,"last_commit_at":77,"category_tags":78,"status":45},1381,"everything-claude-code","affaan-m\u002Feverything-claude-code","everything-claude-code 是一套专为 AI 编程助手（如 Claude Code、Codex、Cursor 等）打造的高性能优化系统。它不仅仅是一组配置文件，而是一个经过长期实战打磨的完整框架，旨在解决 AI 代理在实际开发中面临的效率低下、记忆丢失、安全隐患及缺乏持续学习能力等核心痛点。\n\n通过引入技能模块化、直觉增强、记忆持久化机制以及内置的安全扫描功能，everything-claude-code 能显著提升 AI 在复杂任务中的表现，帮助开发者构建更稳定、更智能的生产级 AI 代理。其独特的“研究优先”开发理念和针对 Token 消耗的优化策略，使得模型响应更快、成本更低，同时有效防御潜在的攻击向量。\n\n这套工具特别适合软件开发者、AI 研究人员以及希望深度定制 AI 工作流的技术团队使用。无论您是在构建大型代码库，还是需要 AI 协助进行安全审计与自动化测试，everything-claude-code 都能提供强大的底层支持。作为一个曾荣获 Anthropic 黑客大奖的开源项目，它融合了多语言支持与丰富的实战钩子（hooks），让 AI 真正成长为懂上",151314,"2026-04-11T23:32:58",[60,35,36],{"id":80,"name":81,"github_repo":82,"description_zh":83,"stars":84,"difficulty_score":27,"last_commit_at":85,"category_tags":86,"status":45},2271,"ComfyUI","Comfy-Org\u002FComfyUI","ComfyUI 是一款功能强大且高度模块化的视觉 AI 引擎，专为设计和执行复杂的 Stable Diffusion 图像生成流程而打造。它摒弃了传统的代码编写模式，采用直观的节点式流程图界面，让用户通过连接不同的功能模块即可构建个性化的生成管线。\n\n这一设计巧妙解决了高级 AI 绘图工作流配置复杂、灵活性不足的痛点。用户无需具备编程背景，也能自由组合模型、调整参数并实时预览效果，轻松实现从基础文生图到多步骤高清修复等各类复杂任务。ComfyUI 拥有极佳的兼容性，不仅支持 Windows、macOS 和 Linux 全平台，还广泛适配 NVIDIA、AMD、Intel 及苹果 Silicon 等多种硬件架构，并率先支持 SDXL、Flux、SD3 等前沿模型。\n\n无论是希望深入探索算法潜力的研究人员和开发者，还是追求极致创作自由度的设计师与资深 AI 绘画爱好者，ComfyUI 都能提供强大的支持。其独特的模块化架构允许社区不断扩展新功能，使其成为当前最灵活、生态最丰富的开源扩散模型工具之一，帮助用户将创意高效转化为现实。",108322,"2026-04-10T11:39:34",[60,61,35],{"id":88,"name":89,"github_repo":90,"description_zh":91,"stars":92,"difficulty_score":27,"last_commit_at":93,"category_tags":94,"status":45},6121,"gemini-cli","google-gemini\u002Fgemini-cli","gemini-cli 是一款由谷歌推出的开源 AI 命令行工具，它将强大的 Gemini 大模型能力直接集成到用户的终端环境中。对于习惯在命令行工作的开发者而言，它提供了一条从输入提示词到获取模型响应的最短路径，无需切换窗口即可享受智能辅助。\n\n这款工具主要解决了开发过程中频繁上下文切换的痛点，让用户能在熟悉的终端界面内直接完成代码理解、生成、调试以及自动化运维任务。无论是查询大型代码库、根据草图生成应用，还是执行复杂的 Git 操作，gemini-cli 都能通过自然语言指令高效处理。\n\n它特别适合广大软件工程师、DevOps 人员及技术研究人员使用。其核心亮点包括支持高达 100 万 token 的超长上下文窗口，具备出色的逻辑推理能力；内置 Google 搜索、文件操作及 Shell 命令执行等实用工具；更独特的是，它支持 MCP（模型上下文协议），允许用户灵活扩展自定义集成，连接如图像生成等外部能力。此外，个人谷歌账号即可享受免费的额度支持，且项目基于 Apache 2.0 协议完全开源，是提升终端工作效率的理想助手。",100752,"2026-04-10T01:20:03",[95,35,61,60],"插件",{"id":97,"name":98,"github_repo":99,"description_zh":100,"stars":101,"difficulty_score":57,"last_commit_at":102,"category_tags":103,"status":45},4487,"LLMs-from-scratch","rasbt\u002FLLMs-from-scratch","LLMs-from-scratch 是一个基于 PyTorch 的开源教育项目，旨在引导用户从零开始一步步构建一个类似 ChatGPT 的大型语言模型（LLM）。它不仅是同名技术著作的官方代码库，更提供了一套完整的实践方案，涵盖模型开发、预训练及微调的全过程。\n\n该项目主要解决了大模型领域“黑盒化”的学习痛点。许多开发者虽能调用现成模型，却难以深入理解其内部架构与训练机制。通过亲手编写每一行核心代码，用户能够透彻掌握 Transformer 架构、注意力机制等关键原理，从而真正理解大模型是如何“思考”的。此外，项目还包含了加载大型预训练权重进行微调的代码，帮助用户将理论知识延伸至实际应用。\n\nLLMs-from-scratch 特别适合希望深入底层原理的 AI 开发者、研究人员以及计算机专业的学生。对于不满足于仅使用 API，而是渴望探究模型构建细节的技术人员而言，这是极佳的学习资源。其独特的技术亮点在于“循序渐进”的教学设计：将复杂的系统工程拆解为清晰的步骤，配合详细的图表与示例，让构建一个虽小但功能完备的大模型变得触手可及。无论你是想夯实理论基础，还是为未来研发更大规模的模型做准备",90106,"2026-04-06T11:19:32",[36,61,35,60]]