docs: 更新 AI 模型版本号 (Claude 4.6, gpt-5.3-codex, Kimi K2.5)
This commit is contained in:
parent
b4e11b037c
commit
00005a305c
34
README.md
34
README.md
|
|
@ -300,16 +300,16 @@ Canvas方式:**代码 ⇄ 白板 ⇄ AI ⇄ 人类**,白板成为单一真
|
|||
|
||||
### AI 模型 & 服务
|
||||
|
||||
* [**Claude Opus 4.5**](https://claude.ai/new): 性能强大的 AI 模型,通过 Claude Code 等平台提供服务,并支持 CLI 和 IDE 插件。
|
||||
* [**gpt-5.1-codex.1-codex (xhigh)**](https://chatgpt.com/codex/): 适用于处理大型项目和复杂逻辑的 AI 模型,可通过 Codex CLI 等平台使用。
|
||||
* [**Droid**](https://factory.ai/news/terminal-bench): 提供对 Claude Opus 4.5 等多种模型的 CLI 访问。
|
||||
* [**Kiro**](https://kiro.dev/): 目前提供免费的 Claude Opus 4.5 模型访问,并提供客户端及 CLI 工具。
|
||||
* [**Claude Opus 4.6**](https://claude.ai/new): 性能强大的 AI 模型,通过 Claude Code 等平台提供服务,并支持 CLI 和 IDE 插件。
|
||||
* [**gpt-5.3-codex (xhigh)**](https://chatgpt.com/codex/): 适用于处理大型项目和复杂逻辑的 AI 模型,可通过 Codex CLI 等平台使用。
|
||||
* [**Droid**](https://factory.ai/news/terminal-bench): 提供对 Claude Opus 4.6 等多种模型的 CLI 访问。
|
||||
* [**Kiro**](https://kiro.dev/): 目前提供免费的 Claude Opus 4.6 模型访问,并提供客户端及 CLI 工具。
|
||||
* [**Gemini CLI**](https://geminicli.com/): 提供对 Gemini 模型的免费访问,适合执行脚本、整理文档和探索思路。
|
||||
* [**antigravity**](https://antigravity.google/): 目前由 Google 提供的免费 AI 服务,支持使用 Claude Opus 4.5 和 Gemini 3.0 Pro。
|
||||
* [**antigravity**](https://antigravity.google/): 目前由 Google 提供的免费 AI 服务,支持使用 Claude Opus 4.6 和 Gemini 3.0 Pro。
|
||||
* [**AI Studio**](https://aistudio.google.com/prompts/new_chat): Google 提供的免费服务,支持使用 Gemini 3.0 Pro 和 Nano Banana。
|
||||
* [**Gemini Enterprise**](https://cloud.google.com/gemini-enterprise): 面向企业用户的 Google AI 服务,目前可以免费使用。
|
||||
* [**GitHub Copilot**](https://github.com/copilot): 由 GitHub 和 OpenAI 联合开发的 AI 代码补全工具。
|
||||
* [**Kimi K2**](https://www.kimi.com/): 一款国产 AI 模型,适用于多种常规任务。
|
||||
* [**Kimi K2.5**](https://www.kimi.com/): 一款国产 AI 模型,适用于多种常规任务。
|
||||
* [**GLM**](https://bigmodel.cn/): 由智谱 AI 开发的国产大语言模型。
|
||||
* [**Qwen**](https://qwenlm.github.io/qwen-code-docs/zh/cli/): 由阿里巴巴开发的 AI 模型,其 CLI 工具提供免费使用额度。
|
||||
|
||||
|
|
@ -589,12 +589,12 @@ gantt
|
|||
> 以下内容翻译自原仓库 [EnzeD/vibe-coding](https://github.com/EnzeD/vibe-coding)
|
||||
|
||||
要开始 Vibe Coding,你只需要以下两种工具之一:
|
||||
- **Claude Opus 4.5**,在 Claude Code 中使用
|
||||
- **gpt-5.1-codex.1-codex (xhigh)**,在 Codex CLI 中使用
|
||||
- **Claude Opus 4.6**,在 Claude Code 中使用
|
||||
- **gpt-5.3-codex (xhigh)**,在 Codex CLI 中使用
|
||||
|
||||
本指南同时适用于 CLI 终端版本和 VSCode 扩展版本(Codex 和 Claude Code 都有扩展,且界面更新)。
|
||||
|
||||
*(注:本指南早期版本使用的是 **Grok 3**,后来切换到 **Gemini 2.5 Pro**,现在我们使用的是 **Claude 4.5**(或 **gpt-5.1-codex.1-codex (xhigh)**))*
|
||||
*(注:本指南早期版本使用的是 **Grok 3**,后来切换到 **Gemini 2.5 Pro**,现在我们使用的是 **Claude 4.5**(或 **gpt-5.3-codex (xhigh)**))*
|
||||
|
||||
*(注2:如果你想使用 Cursor,请查看本指南的 [1.1 版本](https://github.com/EnzeD/vibe-coding/tree/1.1.1),但我们认为它目前不如 Codex CLI 或 Claude Code 强大)*
|
||||
|
||||
|
|
@ -606,14 +606,14 @@ gantt
|
|||
<details>
|
||||
<summary><strong>1. 游戏设计文档(Game Design Document)</strong></summary>
|
||||
|
||||
- 把你的游戏创意交给 **gpt-5.1-codex** 或 **Claude Opus 4.5**,让它生成一份简洁的 **游戏设计文档**,格式为 Markdown,文件名为 `game-design-document.md`。
|
||||
- 把你的游戏创意交给 **gpt-5.3-codex** 或 **Claude Opus 4.6**,让它生成一份简洁的 **游戏设计文档**,格式为 Markdown,文件名为 `game-design-document.md`。
|
||||
- 自己审阅并完善,确保与你的愿景一致。初期可以很简陋,目标是给 AI 提供游戏结构和意图的上下文。不要过度设计,后续会迭代。
|
||||
</details>
|
||||
|
||||
<details>
|
||||
<summary><strong>2. 技术栈与 <code>CLAUDE.md</code> / <code>Agents.md</code></strong></summary>
|
||||
|
||||
- 让 **gpt-5.1-codex** 或 **Claude Opus 4.5** 为你的游戏推荐最合适的技术栈(例如:多人3D游戏用 ThreeJS + WebSocket),保存为 `tech-stack.md`。
|
||||
- 让 **gpt-5.3-codex** 或 **Claude Opus 4.6** 为你的游戏推荐最合适的技术栈(例如:多人3D游戏用 ThreeJS + WebSocket),保存为 `tech-stack.md`。
|
||||
- 要求它提出 **最简单但最健壮** 的技术栈。
|
||||
- 在终端中打开 **Claude Code** 或 **Codex CLI**,使用 `/init` 命令,它会读取你已创建的两个 .md 文件,生成一套规则来正确引导大模型。
|
||||
- **关键:一定要审查生成的规则。** 确保规则强调 **模块化**(多文件)和禁止 **单体巨文件**(monolith)。可能需要手动修改或补充规则。
|
||||
|
|
@ -631,7 +631,7 @@ gantt
|
|||
<details>
|
||||
<summary><strong>3. 实施计划(Implementation Plan)</strong></summary>
|
||||
|
||||
- 将以下内容提供给 **gpt-5.1-codex** 或 **Claude Opus 4.5**:
|
||||
- 将以下内容提供给 **gpt-5.3-codex** 或 **Claude Opus 4.6**:
|
||||
- 游戏设计文档(`game-design-document.md`)
|
||||
- 技术栈推荐(`tech-stack.md`)
|
||||
- 让它生成一份详细的 **实施计划**(Markdown 格式),包含一系列给 AI 开发者的分步指令。
|
||||
|
|
@ -675,7 +675,7 @@ gantt
|
|||
- 打开 **Codex** 或 **Claude Code**(扩展或终端)。
|
||||
- 提示词:阅读 `/memory-bank` 所有文档,然后执行实施计划的第 1 步。我会负责跑测试。在我验证测试通过前,不要开始第 2 步。验证通过后,打开 `progress.md` 记录你做了什么供后续开发者参考,再把新的架构洞察添加到 `architecture.md` 中解释每个文件的作用。
|
||||
- **永远** 先用 "Ask" 模式或 "Plan Mode"(Claude Code 中按 `shift+tab`),确认满意后再让 AI 执行该步骤。
|
||||
- **极致 Vibe:** 安装 [Superwhisper](https://superwhisper.com),用语音随便跟 Claude 或 gpt-5.1-codex 聊天,不用打字。
|
||||
- **极致 Vibe:** 安装 [Superwhisper](https://superwhisper.com),用语音随便跟 Claude 或 gpt-5.3-codex 聊天,不用打字。
|
||||
</details>
|
||||
|
||||
<details>
|
||||
|
|
@ -707,7 +707,7 @@ gantt
|
|||
<summary><strong>常规修复</strong></summary>
|
||||
|
||||
- 如果某个提示词失败或搞崩了项目:
|
||||
- Claude Code 用 `/rewind` 回退;用 gpt-5.1-codex 的话多提交 git,需要时 reset。
|
||||
- Claude Code 用 `/rewind` 回退;用 gpt-5.3-codex 的话多提交 git,需要时 reset。
|
||||
- 报错处理:
|
||||
- **JavaScript 错误:** 打开浏览器控制台(F12),复制错误,贴给 AI;视觉问题截图发给它。
|
||||
- **懒人方案:** 安装 [BrowserTools](https://browsertools.agentdesk.ai/installation),自动复制错误和截图。
|
||||
|
|
@ -719,7 +719,7 @@ gantt
|
|||
- 实在卡住:
|
||||
- 回退到上一个 git commit(`git reset`),换新提示词重试。
|
||||
- 极度卡壳:
|
||||
- 用 [RepoPrompt](https://repoprompt.com/) 或 [uithub](https://uithub.com/) 把整个代码库合成一个文件,然后丢给 **gpt-5.1-codex 或 Claude** 求救。
|
||||
- 用 [RepoPrompt](https://repoprompt.com/) 或 [uithub](https://uithub.com/) 把整个代码库合成一个文件,然后丢给 **gpt-5.3-codex 或 Claude** 求救。
|
||||
</details>
|
||||
|
||||
</details>
|
||||
|
|
@ -740,7 +740,7 @@ gantt
|
|||
<details>
|
||||
<summary><strong>其他实用技巧</strong></summary>
|
||||
|
||||
- **小修改:** 用 gpt-5.1-codex (medium)
|
||||
- **小修改:** 用 gpt-5.3-codex (medium)
|
||||
- **写顶级营销文案:** 用 Opus 4.1
|
||||
- **生成优秀 2D 精灵图:** 用 ChatGPT + Nano Banana
|
||||
- **生成音乐:** 用 Suno
|
||||
|
|
@ -763,7 +763,7 @@ gantt
|
|||
- **A:** 那不是一个提示词,是 ~30 个提示词 + 专门的 `plane-implementation.md` 文件引导的。用精准指令如“在机翼上为副翼切出空间”,而不是“做一个飞机”这种模糊指令。
|
||||
|
||||
- **Q: 为什么现在 Claude Code 或 Codex CLI 比 Cursor 更强?**
|
||||
- **A:** 完全看个人喜好。我们强调的是:Claude Code 能更好发挥 Claude Opus 4.5 的实力,Codex CLI 能更好发挥 gpt-5.1-codex 的实力,而 Cursor 对这两者的利用都不如原生终端版。终端版还能在任意 IDE、使用 SSH 远程服务器等场景工作,自定义命令、子代理、钩子等功能也能长期大幅提升开发质量和速度。最后,即使你只是低配 Claude 或 ChatGPT 订阅,也完全够用。
|
||||
- **A:** 完全看个人喜好。我们强调的是:Claude Code 能更好发挥 Claude Opus 4.6 的实力,Codex CLI 能更好发挥 gpt-5.3-codex 的实力,而 Cursor 对这两者的利用都不如原生终端版。终端版还能在任意 IDE、使用 SSH 远程服务器等场景工作,自定义命令、子代理、钩子等功能也能长期大幅提升开发质量和速度。最后,即使你只是低配 Claude 或 ChatGPT 订阅,也完全够用。
|
||||
|
||||
- **Q: 我不会搭建多人游戏的服务器怎么办?**
|
||||
- **A:** 问你的 AI。
|
||||
|
|
|
|||
|
|
@ -299,16 +299,16 @@ Through this continuous **recursive optimization loop**, the system achieves **s
|
|||
|
||||
### AI Models & Services
|
||||
|
||||
* [**Claude Opus 4.5**](https://claude.ai/new): A powerful AI model, offered through platforms like Claude Code, and supporting CLI and IDE plugins.
|
||||
* [**gpt-5.1-codex.1-codex (xhigh)**](https://chatgpt.com/codex/): An AI model suitable for handling large projects and complex logic, usable through platforms like Codex CLI.
|
||||
* [**Droid**](https://factory.ai/news/terminal-bench): Provides CLI access to various models including Claude Opus 4.5.
|
||||
* [**Kiro**](https://kiro.dev/): Currently offers free access to the Claude Opus 4.5 model, and provides client and CLI tools.
|
||||
* [**Claude Opus 4.6**](https://claude.ai/new): A powerful AI model, offered through platforms like Claude Code, and supporting CLI and IDE plugins.
|
||||
* [**gpt-5.3-codex (xhigh)**](https://chatgpt.com/codex/): An AI model suitable for handling large projects and complex logic, usable through platforms like Codex CLI.
|
||||
* [**Droid**](https://factory.ai/news/terminal-bench): Provides CLI access to various models including Claude Opus 4.6.
|
||||
* [**Kiro**](https://kiro.dev/): Currently offers free access to the Claude Opus 4.6 model, and provides client and CLI tools.
|
||||
* [**Gemini CLI**](https://geminicli.com/): Provides free access to the Gemini model, suitable for executing scripts, organizing documents, and exploring ideas.
|
||||
* [**antigravity**](https://antigravity.google/): Currently a free AI service provided by Google, supporting Claude Opus 4.5 and Gemini 3.0 Pro.
|
||||
* [**antigravity**](https://antigravity.google/): Currently a free AI service provided by Google, supporting Claude Opus 4.6 and Gemini 3.0 Pro.
|
||||
* [**AI Studio**](https://aistudio.google.com/prompts/new_chat): A free service provided by Google, supporting Gemini 3.0 Pro and Nano Banana.
|
||||
* [**Gemini Enterprise**](https://cloud.google.com/gemini-enterprise): Google's AI service for enterprise users, currently available for free.
|
||||
* [**GitHub Copilot**](https://github.com/copilot): An AI code completion tool jointly developed by GitHub and OpenAI.
|
||||
* [**Kimi K2**](https://www.kimi.com/): A domestic AI model suitable for various general tasks.
|
||||
* [**Kimi K2.5**](https://www.kimi.com/): A domestic AI model suitable for various general tasks.
|
||||
* [**GLM**](https://bigmodel.cn/): A domestic large language model developed by Zhipu AI.
|
||||
* [**Qwen**](https://qwenlm.github.io/qwen-code-docs/zh/cli/): An AI model developed by Alibaba, its CLI tool offers free usage quota.
|
||||
|
||||
|
|
@ -587,12 +587,12 @@ gantt
|
|||
> The following content is translated from the original repository [EnzeD/vibe-coding](https://github.com/EnzeD/vibe-coding)
|
||||
|
||||
To start Vibe Coding, you only need one of the following two tools:
|
||||
- **Claude Opus 4.5**, used in Claude Code
|
||||
- **gpt-5.1-codex.1-codex (xhigh)**, used in Codex CLI
|
||||
- **Claude Opus 4.6**, used in Claude Code
|
||||
- **gpt-5.3-codex (xhigh)**, used in Codex CLI
|
||||
|
||||
This guide applies to both the CLI terminal version and the VSCode extension version (both Codex and Claude Code have extensions, and their interfaces are updated).
|
||||
|
||||
*(Note: Earlier versions of this guide used **Grok 3**, later switched to **Gemini 2.5 Pro**, and now we are using **Claude 4.5** (or **gpt-5.1-codex.1-codex (xhigh)**))*
|
||||
*(Note: Earlier versions of this guide used **Grok 3**, later switched to **Gemini 2.5 Pro**, and now we are using **Claude 4.5** (or **gpt-5.3-codex (xhigh)**))*
|
||||
|
||||
*(Note 2: If you want to use Cursor, please check version [1.1](https://github.com/EnzeD/vibe-coding/tree/1.1.1) of this guide, but we believe it is currently less powerful than Codex CLI or Claude Code)*
|
||||
|
||||
|
|
@ -604,14 +604,14 @@ This guide applies to both the CLI terminal version and the VSCode extension ver
|
|||
<details>
|
||||
<summary><strong>1. Game Design Document</strong></summary>
|
||||
|
||||
- Hand your game idea to **gpt-5.1-codex** or **Claude Opus 4.5** to generate a concise **Game Design Document** in Markdown format, named `game-design-document.md`.
|
||||
- Hand your game idea to **gpt-5.3-codex** or **Claude Opus 4.6** to generate a concise **Game Design Document** in Markdown format, named `game-design-document.md`.
|
||||
- Review and refine it yourself to ensure it aligns with your vision. It can be very basic initially; the goal is to provide AI with the game structure and intent context. Do not over-design; it will be iterated later.
|
||||
</details>
|
||||
|
||||
<details>
|
||||
<summary><strong>2. Tech Stack and <code>CLAUDE.md</code> / <code>Agents.md</code></strong></summary>
|
||||
|
||||
- Ask **gpt-5.1-codex** or **Claude Opus 4.5** to recommend the most suitable tech stack for your game (e.g., ThreeJS + WebSocket for a multiplayer 3D game), save it as `tech-stack.md`.
|
||||
- Ask **gpt-5.3-codex** or **Claude Opus 4.6** to recommend the most suitable tech stack for your game (e.g., ThreeJS + WebSocket for a multiplayer 3D game), save it as `tech-stack.md`.
|
||||
- Ask it to propose the **simplest yet most robust** tech stack.
|
||||
- Open **Claude Code** or **Codex CLI** in your terminal and use the `/init` command. It will read the two `.md` files you've created and generate a set of rules to guide the large model correctly.
|
||||
- **Key: Always review the generated rules.** Ensure the rules emphasize **modularization** (multiple files) and prohibit **monolithic files**. You may need to manually modify or supplement the rules.
|
||||
|
|
@ -629,7 +629,7 @@ This guide applies to both the CLI terminal version and the VSCode extension ver
|
|||
<details>
|
||||
<summary><strong>3. Implementation Plan</strong></summary>
|
||||
|
||||
- Provide the following to **gpt-5.1-codex** or **Claude Opus 4.5**:
|
||||
- Provide the following to **gpt-5.3-codex** or **Claude Opus 4.6**:
|
||||
- Game Design Document (`game-design-document.md`)
|
||||
- Tech Stack Recommendation (`tech-stack.md`)
|
||||
- Ask it to generate a detailed **Implementation Plan** (Markdown format), containing a series of step-by-step instructions for AI developers.
|
||||
|
|
@ -673,7 +673,7 @@ Now for the most exciting part!
|
|||
- Open **Codex** or **Claude Code** (extension or terminal).
|
||||
- Prompt: Read all documents in `/memory-bank`, then execute step 1 of the implementation plan. I will be responsible for running tests. Do not start step 2 until I verify the tests pass. After verification, open `progress.md` to record what you've done for future developers' reference, and add new architectural insights to `architecture.md` explaining the purpose of each file.
|
||||
- **Always** use "Ask" mode or "Plan Mode" (press `shift+tab` in Claude Code) first, and only let AI execute the step after you are satisfied.
|
||||
- **Ultimate Vibe:** Install [Superwhisper](https://superwhisper.com) and chat casually with Claude or gpt-5.1-codex using voice, without typing.
|
||||
- **Ultimate Vibe:** Install [Superwhisper](https://superwhisper.com) and chat casually with Claude or gpt-5.3-codex using voice, without typing.
|
||||
</details>
|
||||
|
||||
<details>
|
||||
|
|
@ -705,7 +705,7 @@ Congratulations! You've built a basic game! It might still be rough and lack fea
|
|||
<summary><strong>General Fixes</strong></summary>
|
||||
|
||||
- If a prompt fails or breaks the project:
|
||||
- Use `/rewind` in Claude Code to revert; for gpt-5.1-codex, commit frequently with Git and reset when needed.
|
||||
- Use `/rewind` in Claude Code to revert; for gpt-5.3-codex, commit frequently with Git and reset when needed.
|
||||
- Error handling:
|
||||
- **JavaScript errors:** Open browser console (F12), copy error, paste to AI; for visual issues, send a screenshot.
|
||||
- **Lazy solution:** Install [BrowserTools](https://browsertools.agentdesk.ai/installation) to automatically copy errors and screenshots.
|
||||
|
|
@ -717,7 +717,7 @@ Congratulations! You've built a basic game! It might still be rough and lack fea
|
|||
- Really stuck:
|
||||
- Revert to the previous git commit (`git reset`), try again with a new prompt.
|
||||
- Extremely stuck:
|
||||
- Use [RepoPrompt](https://repoprompt.com/) or [uithub](https://uithub.com/) to synthesize the entire codebase into one file, then send it to **gpt-5.1-codex or Claude** for help.
|
||||
- Use [RepoPrompt](https://repoprompt.com/) or [uithub](https://uithub.com/) to synthesize the entire codebase into one file, then send it to **gpt-5.3-codex or Claude** for help.
|
||||
</details>
|
||||
|
||||
</details>
|
||||
|
|
@ -738,7 +738,7 @@ Congratulations! You've built a basic game! It might still be rough and lack fea
|
|||
<details>
|
||||
<summary><strong>Other Useful Tips</strong></summary>
|
||||
|
||||
- **Small modifications:** Use gpt-5.1-codex (medium)
|
||||
- **Small modifications:** Use gpt-5.3-codex (medium)
|
||||
- **Write top-tier marketing copy:** Use Opus 4.1
|
||||
- **Generate excellent 2D sprites:** Use ChatGPT + Nano Banana
|
||||
- **Generate music:** Use Suno
|
||||
|
|
@ -761,7 +761,7 @@ Congratulations! You've built a basic game! It might still be rough and lack fea
|
|||
- **A:** That wasn't one prompt, it was ~30 prompts + a dedicated `plane-implementation.md` file guided it. Use precise instructions like "cut space for ailerons on the wing," instead of vague instructions like "make a plane."
|
||||
|
||||
- **Q: Why are Claude Code or Codex CLI stronger than Cursor now?**
|
||||
- **A:** It's entirely a matter of personal preference. We emphasize that Claude Code can better leverage the power of Claude Opus 4.5, and Codex CLI can better leverage the power of gpt-5.1-codex. Cursor does not utilize either of these as well as their native terminal versions. Terminal versions can also work in any IDE, with SSH remote servers, etc., and features like custom commands, sub-agents, and hooks can significantly improve development quality and speed in the long run. Finally, even if you only have a low-tier Claude or ChatGPT subscription, it's completely sufficient.
|
||||
- **A:** It's entirely a matter of personal preference. We emphasize that Claude Code can better leverage the power of Claude Opus 4.6, and Codex CLI can better leverage the power of gpt-5.3-codex. Cursor does not utilize either of these as well as their native terminal versions. Terminal versions can also work in any IDE, with SSH remote servers, etc., and features like custom commands, sub-agents, and hooks can significantly improve development quality and speed in the long run. Finally, even if you only have a low-tier Claude or ChatGPT subscription, it's completely sufficient.
|
||||
|
||||
- **Q: What if I don't know how to set up a multiplayer game server?**
|
||||
- **A:** Ask your AI.
|
||||
|
|
|
|||
|
|
@ -125,7 +125,7 @@ AI CLI tools typically ask for confirmation by default; enabling full permission
|
|||
|
||||
```bash
|
||||
# Codex - Most powerful configuration
|
||||
codex --enable web_search_request -m gpt-5.1-codex-max -c model_reasoning_effort="high" --dangerously-bypass-approvals-and-sandbox
|
||||
codex --enable web_search_request -m gpt-5.3-codex-max -c model_reasoning_effort="high" --dangerously-bypass-approvals-and-sandbox
|
||||
|
||||
# Claude Code - Skip all confirmations
|
||||
claude --dangerously-skip-permissions
|
||||
|
|
@ -140,7 +140,7 @@ Add the following configuration to `~/.bashrc` to launch AI with a single letter
|
|||
|
||||
```bash
|
||||
# c - Codex (GPT-5.1 most powerful mode)
|
||||
alias c='codex --enable web_search_request -m gpt-5.1-codex-max -c model_reasoning_effort="high" --dangerously-bypass-approvals-and-sandbox'
|
||||
alias c='codex --enable web_search_request -m gpt-5.3-codex-max -c model_reasoning_effort="high" --dangerously-bypass-approvals-and-sandbox'
|
||||
|
||||
# cc - Claude Code (full permissions)
|
||||
alias cc='claude --dangerously-skip-permissions'
|
||||
|
|
|
|||
|
|
@ -45,7 +45,7 @@
|
|||
|
||||
| Platform | Model | Features |
|
||||
|:---|:---|:---|
|
||||
| [Claude](https://claude.ai/) | Claude Opus 4.5 | Strong coding ability, supports Artifacts |
|
||||
| [Claude](https://claude.ai/) | Claude Opus 4.6 | Strong coding ability, supports Artifacts |
|
||||
| [ChatGPT](https://chatgpt.com/) | GPT-5.1 | Strong comprehensive ability, supports Codex |
|
||||
| [Gemini](https://gemini.google.com/) | Gemini 3.0 Pro | Large free quota, supports long context |
|
||||
|
||||
|
|
@ -53,7 +53,7 @@
|
|||
|
||||
| Platform | Model | Features |
|
||||
|:---|:---|:---|
|
||||
| [Kimi](https://kimi.moonshot.cn/) | Kimi K2 | Strong long text processing |
|
||||
| [Kimi](https://kimi.moonshot.cn/) | Kimi K2.5 | Strong long text processing |
|
||||
| [Tongyi Qianwen](https://tongyi.aliyun.com/) | Qwen | By Alibaba, free |
|
||||
| [Zhipu Qingyan](https://chatglm.cn/) | GLM-4 | By Zhipu AI |
|
||||
| [Doubao](https://www.doubao.com/) | Doubao | By ByteDance |
|
||||
|
|
@ -103,7 +103,7 @@
|
|||
|
||||
| Resource | Link | Description |
|
||||
|:---|:---|:---|
|
||||
| Kiro | [kiro.dev](https://kiro.dev/) | Free Claude Opus 4.5 |
|
||||
| Kiro | [kiro.dev](https://kiro.dev/) | Free Claude Opus 4.6 |
|
||||
| Windsurf | [windsurf.com](https://windsurf.com/) | Free quota for new users |
|
||||
| GitHub Copilot | [github.com/copilot](https://github.com/copilot) | Free for students/open-source |
|
||||
| Codeium | [codeium.com](https://codeium.com/) | Free AI code completion |
|
||||
|
|
|
|||
|
|
@ -18,7 +18,7 @@
|
|||
|
||||
| Model | Description |
|
||||
|:---|:---|
|
||||
| Claude Opus 4.5 | Strong code capabilities |
|
||||
| Claude Opus 4.6 | Strong code capabilities |
|
||||
| GPT-5.1 Codex | Complex logic processing |
|
||||
| Gemini 2.5 Pro | Free long context |
|
||||
|
||||
|
|
|
|||
|
|
@ -83,12 +83,12 @@
|
|||
|
||||
## 📋 The Tools (Qi)
|
||||
|
||||
- [**Claude Opus 4.5**](https://claude.ai/new), used in Claude Code, very expensive, but iOS subscription in some regions is hundreds of RMB cheaper, fast + good effect, top tier, has CLI and IDE plugins.
|
||||
- [**gpt-5.1-codex.1-codex (xhigh)**](https://chatgpt.com/codex/), used in Codex CLI, top tier, nothing to complain about except being slow, the only solution for large projects with complex logic, available with ChatGPT membership, has CLI and IDE plugins.
|
||||
- [**Droid**](https://factory.ai/news/terminal-bench), the Claude Opus 4.5 here is even stronger than Claude Code, top tier, has CLI.
|
||||
- [**Kiro**](https://kiro.dev/), the Claude Opus 4.5 here is currently free, but the CLI is a bit weak, can't see the running status, has client and CLI.
|
||||
- [**Claude Opus 4.6**](https://claude.ai/new), used in Claude Code, very expensive, but iOS subscription in some regions is hundreds of RMB cheaper, fast + good effect, top tier, has CLI and IDE plugins.
|
||||
- [**gpt-5.3-codex (xhigh)**](https://chatgpt.com/codex/), used in Codex CLI, top tier, nothing to complain about except being slow, the only solution for large projects with complex logic, available with ChatGPT membership, has CLI and IDE plugins.
|
||||
- [**Droid**](https://factory.ai/news/terminal-bench), the Claude Opus 4.6 here is even stronger than Claude Code, top tier, has CLI.
|
||||
- [**Kiro**](https://kiro.dev/), the Claude Opus 4.6 here is currently free, but the CLI is a bit weak, can't see the running status, has client and CLI.
|
||||
- [**gemini**](https://geminicli.com/), currently free to use, for dirty work, can execute scripts written by Claude Code or Codex, also good for organizing documents and brainstorming, has client and CLI.
|
||||
- [**antigravity**](https://antigravity.google/), Google's, free to use Claude Opus 4.5 and Gemini 3.0 Pro, a great philanthropist.
|
||||
- [**antigravity**](https://antigravity.google/), Google's, free to use Claude Opus 4.6 and Gemini 3.0 Pro, a great philanthropist.
|
||||
- [**aistudio**](https://aistudio.google.com/prompts/new_chat), from Google, free to use Gemini 3.0 Pro and Nano Banana.
|
||||
- [**gemini-enterprise**](https://cloud.google.com/gemini-enterprise), Google enterprise version, currently free to use Nano Banana Pro.
|
||||
- [**augment**](https://app.augmentcode.com/), its context engine and prompt optimization button are simply divine, beginners should just use it, click the button to automatically write prompts for you, a must-have for lazy people.
|
||||
|
|
@ -370,12 +370,12 @@ gantt
|
|||
|
||||
## 🚀 Getting Started (This is by the original author, not me, I updated what I think are the best models)
|
||||
To start Vibe Coding, you only need one of the following two tools:
|
||||
- **Claude Opus 4.5**, used in Claude Code
|
||||
- **gpt-5.1-codex.1-codex (xhigh)**, used in Codex CLI
|
||||
- **Claude Opus 4.6**, used in Claude Code
|
||||
- **gpt-5.3-codex (xhigh)**, used in Codex CLI
|
||||
|
||||
This guide applies to both the CLI terminal version and the VSCode extension version (both Codex and Claude Code have extensions, and their interfaces are updated).
|
||||
|
||||
*(Note: Earlier versions of this guide used **Grok 3**, later switched to **Gemini 2.5 Pro**, and now we are using **Claude 4.5** (or **gpt-5.1-codex.1-codex (xhigh)**))*
|
||||
*(Note: Earlier versions of this guide used **Grok 3**, later switched to **Gemini 2.5 Pro**, and now we are using **Claude 4.5** (or **gpt-5.3-codex (xhigh)**))*
|
||||
|
||||
*(Note 2: If you want to use Cursor, please check version [1.1](https://github.com/EnzeD/vibe-coding/tree/1.1.1) of this guide, but we believe it is currently less powerful than Codex CLI or Claude Code)*
|
||||
|
||||
|
|
@ -387,14 +387,14 @@ This guide applies to both the CLI terminal version and the VSCode extension ver
|
|||
<details>
|
||||
<summary><strong>1. Game Design Document</strong></summary>
|
||||
|
||||
- Hand your game idea to **gpt-5.1-codex** or **Claude Opus 4.5** to generate a concise **Game Design Document** in Markdown format, named `game-design-document.md`.
|
||||
- Hand your game idea to **gpt-5.3-codex** or **Claude Opus 4.6** to generate a concise **Game Design Document** in Markdown format, named `game-design-document.md`.
|
||||
- Review and refine it yourself to ensure it aligns with your vision. It can be very basic initially; the goal is to provide AI with the game structure and intent context. Do not over-design; it will be iterated later.
|
||||
</details>
|
||||
|
||||
<details>
|
||||
<summary><strong>2. Tech Stack and <code>CLAUDE.md</code> / <code>Agents.md</code></strong></summary>
|
||||
|
||||
- Ask **gpt-5.1-codex** or **Claude Opus 4.5** to recommend the most suitable tech stack for your game (e.g., ThreeJS + WebSocket for a multiplayer 3D game), save it as `tech-stack.md`.
|
||||
- Ask **gpt-5.3-codex** or **Claude Opus 4.6** to recommend the most suitable tech stack for your game (e.g., ThreeJS + WebSocket for a multiplayer 3D game), save it as `tech-stack.md`.
|
||||
- Ask it to propose the **simplest yet most robust** tech stack.
|
||||
- Open **Claude Code** or **Codex CLI** in your terminal and use the `/init` command. It will read the two `.md` files you've created and generate a set of rules to guide the large model correctly.
|
||||
- **Key: Always review the generated rules.** Ensure the rules emphasize **modularization** (multiple files) and prohibit **monolithic files**. You may need to manually modify or supplement the rules.
|
||||
|
|
@ -412,7 +412,7 @@ This guide applies to both the CLI terminal version and the VSCode extension ver
|
|||
<details>
|
||||
<summary><strong>3. Implementation Plan</strong></summary>
|
||||
|
||||
- Provide the following to **gpt-5.1-codex** or **Claude Opus 4.5**:
|
||||
- Provide the following to **gpt-5.3-codex** or **Claude Opus 4.6**:
|
||||
- Game Design Document (`game-design-document.md`)
|
||||
- Tech Stack Recommendation (`tech-stack.md`)
|
||||
- Ask it to generate a detailed **Implementation Plan** (Markdown format), containing a series of step-by-step instructions for AI developers.
|
||||
|
|
@ -456,7 +456,7 @@ Now for the most exciting part!
|
|||
- Open **Codex** or **Claude Code** (extension or terminal).
|
||||
- Prompt: Read all documents in `/memory-bank`, then execute step 1 of the implementation plan. I will be responsible for running tests. Do not start step 2 until I verify the tests pass. After verification, open `progress.md` to record what you've done for future developers' reference, and add new architectural insights to `architecture.md` explaining the purpose of each file.
|
||||
- **Always** use "Ask" mode or "Plan Mode" (press `shift+tab` in Claude Code) first, and only let AI execute the step after you are satisfied.
|
||||
- **Ultimate Vibe:** Install [Superwhisper](https://superwhisper.com) and chat casually with Claude or gpt-5.1-codex using voice, without typing.
|
||||
- **Ultimate Vibe:** Install [Superwhisper](https://superwhisper.com) and chat casually with Claude or gpt-5.3-codex using voice, without typing.
|
||||
</details>
|
||||
|
||||
<details>
|
||||
|
|
@ -488,7 +488,7 @@ Congratulations! You've built a basic game! It might still be rough and lack fea
|
|||
<summary><strong>General Fixes</strong></summary>
|
||||
|
||||
- If a prompt fails or breaks the project:
|
||||
- Use `/rewind` in Claude Code to revert; for gpt-5.1-codex, commit frequently with Git and reset when needed.
|
||||
- Use `/rewind` in Claude Code to revert; for gpt-5.3-codex, commit frequently with Git and reset when needed.
|
||||
- Error handling:
|
||||
- **JavaScript errors:** Open browser console (F12), copy error, paste to AI; for visual issues, send a screenshot.
|
||||
- **Lazy solution:** Install [BrowserTools](https://browsertools.agentdesk.ai/installation) to automatically copy errors and screenshots.
|
||||
|
|
@ -500,7 +500,7 @@ Congratulations! You've built a basic game! It might still be rough and lack fea
|
|||
- Really stuck:
|
||||
- Revert to the previous git commit (`git reset`), try again with a new prompt.
|
||||
- Extremely stuck:
|
||||
- Use [RepoPrompt](https://repoprompt.com/) or [uithub](https://uithub.com/) to synthesize the entire codebase into one file, then send it to **gpt-5.1-codex or Claude** for help.
|
||||
- Use [RepoPrompt](https://repoprompt.com/) or [uithub](https://uithub.com/) to synthesize the entire codebase into one file, then send it to **gpt-5.3-codex or Claude** for help.
|
||||
</details>
|
||||
|
||||
</details>
|
||||
|
|
@ -521,7 +521,7 @@ Congratulations! You've built a basic game! It might still be rough and lack fea
|
|||
<details>
|
||||
<summary><strong>Other Useful Tips</strong></summary>
|
||||
|
||||
- **Small modifications:** Use gpt-5.1-codex (medium)
|
||||
- **Small modifications:** Use gpt-5.3-codex (medium)
|
||||
- **Write top-tier marketing copy:** Use Opus 4.1
|
||||
- **Generate excellent 2D sprites:** Use ChatGPT + Nano Banana
|
||||
- **Generate music:** Use Suno
|
||||
|
|
@ -544,7 +544,7 @@ Congratulations! You've built a basic game! It might still be rough and lack fea
|
|||
- **A:** That wasn't one prompt, it was ~30 prompts + a dedicated `plane-implementation.md` file guided it. Use precise instructions like "cut space for ailerons on the wing," instead of vague instructions like "make a plane."
|
||||
|
||||
- **Q: Why are Claude Code or Codex CLI stronger than Cursor now?**
|
||||
- **A:** It's entirely a matter of personal preference. We emphasize that Claude Code can better leverage the power of Claude Opus 4.5, and Codex CLI can better leverage the power of gpt-5.1-codex. Cursor does not utilize either of these as well as their native terminal versions. Terminal versions can also work in any IDE, with SSH remote servers, etc., and features like custom commands, sub-agents, and hooks can significantly improve development quality and speed in the long run. Finally, even if you only have a low-tier Claude or ChatGPT subscription, it's completely sufficient.
|
||||
- **A:** It's entirely a matter of personal preference. We emphasize that Claude Code can better leverage the power of Claude Opus 4.6, and Codex CLI can better leverage the power of gpt-5.3-codex. Cursor does not utilize either of these as well as their native terminal versions. Terminal versions can also work in any IDE, with SSH remote servers, etc., and features like custom commands, sub-agents, and hooks can significantly improve development quality and speed in the long run. Finally, even if you only have a low-tier Claude or ChatGPT subscription, it's completely sufficient.
|
||||
|
||||
- **Q: What if I don't know how to set up a multiplayer game server?**
|
||||
- **A:** Ask your AI.
|
||||
|
|
|
|||
|
|
@ -84,12 +84,12 @@ TRANSLATED CONTENT:
|
|||
|
||||
## 📋 器
|
||||
|
||||
- [**Claude Opus 4.5**](https://claude.ai/new),在 Claude Code 中使用 很贵,但是尼区ios订阅要便宜几百人民币,快+效果好,顶中顶中顶,有 cli 和 ide 插件
|
||||
- [**gpt-5.1-codex.1-codex (xhigh)**](https://chatgpt.com/codex/),在 Codex CLI 中使用,顶中顶,除了慢其他没得挑,大项目复杂逻辑唯一解,买chatgpt会员就能用,有 cli 和 ide 插件
|
||||
- [**Droid**](https://factory.ai/news/terminal-bench),这个里面的 Claude Opus 4.5比 Claude Code 还强,顶,有 cli
|
||||
- [**Kiro**](https://kiro.dev/),这个里面的 Claude Opus 4.5 现在免费,就是cli有点拉,看不到正在运行的情况有客户端和 cli
|
||||
- [**Claude Opus 4.6**](https://claude.ai/new),在 Claude Code 中使用 很贵,但是尼区ios订阅要便宜几百人民币,快+效果好,顶中顶中顶,有 cli 和 ide 插件
|
||||
- [**gpt-5.3-codex (xhigh)**](https://chatgpt.com/codex/),在 Codex CLI 中使用,顶中顶,除了慢其他没得挑,大项目复杂逻辑唯一解,买chatgpt会员就能用,有 cli 和 ide 插件
|
||||
- [**Droid**](https://factory.ai/news/terminal-bench),这个里面的 Claude Opus 4.6比 Claude Code 还强,顶,有 cli
|
||||
- [**Kiro**](https://kiro.dev/),这个里面的 Claude Opus 4.6 现在免费,就是cli有点拉,看不到正在运行的情况有客户端和 cli
|
||||
- [**gemini**](https://geminicli.com/),目前免费用,干脏活,用 Claude Code 或者 codex 写好的脚本,拿他来执行可以,整理文档和找思路就它了有客户端和 cli
|
||||
- [**antigravity**](https://antigravity.google/),谷歌的,可以免费用 Claude Opus 4.5 和 gemini 3.0 pro 大善人
|
||||
- [**antigravity**](https://antigravity.google/),谷歌的,可以免费用 Claude Opus 4.6 和 gemini 3.0 pro 大善人
|
||||
- [**aistudio**](https://aistudio.google.com/prompts/new_chat),谷歌家的,免费用 gemini 3.0 pro 和 Nano Banana
|
||||
- [**gemini-enterprise**](https://cloud.google.com/gemini-enterprise),谷歌企业版,现在能免费用 Nano Banana pro
|
||||
- [**augment**](https://app.augmentcode.com/),它的上下文引擎和提示词优化按钮真的神中神中神,小白就用它就行了,点击按钮自动帮你写好提示词,懒人必备
|
||||
|
|
@ -371,12 +371,12 @@ gantt
|
|||
|
||||
## 🚀 入门指南(这里是原作者的,不是我写的,我更新了一下我认为最好的模型)
|
||||
要开始 Vibe Coding,你只需要以下两种工具之一:
|
||||
- **Claude Opus 4.5**,在 Claude Code 中使用
|
||||
- **gpt-5.1-codex.1-codex (xhigh)**,在 Codex CLI 中使用
|
||||
- **Claude Opus 4.6**,在 Claude Code 中使用
|
||||
- **gpt-5.3-codex (xhigh)**,在 Codex CLI 中使用
|
||||
|
||||
本指南同时适用于 CLI 终端版本和 VSCode 扩展版本(Codex 和 Claude Code 都有扩展,且界面更新)。
|
||||
|
||||
*(注:本指南早期版本使用的是 **Grok 3**,后来切换到 **Gemini 2.5 Pro**,现在我们使用的是 **Claude 4.5**(或 **gpt-5.1-codex.1-codex (xhigh)**))*
|
||||
*(注:本指南早期版本使用的是 **Grok 3**,后来切换到 **Gemini 2.5 Pro**,现在我们使用的是 **Claude 4.5**(或 **gpt-5.3-codex (xhigh)**))*
|
||||
|
||||
*(注2:如果你想使用 Cursor,请查看本指南的 [1.1 版本](https://github.com/EnzeD/vibe-coding/tree/1.1.1),但我们认为它目前不如 Codex CLI 或 Claude Code 强大)*
|
||||
|
||||
|
|
@ -388,14 +388,14 @@ gantt
|
|||
<details>
|
||||
<summary><strong>1. 游戏设计文档(Game Design Document)</strong></summary>
|
||||
|
||||
- 把你的游戏创意交给 **gpt-5.1-codex** 或 **Claude Opus 4.5**,让它生成一份简洁的 **游戏设计文档**,格式为 Markdown,文件名为 `game-design-document.md`。
|
||||
- 把你的游戏创意交给 **gpt-5.3-codex** 或 **Claude Opus 4.6**,让它生成一份简洁的 **游戏设计文档**,格式为 Markdown,文件名为 `game-design-document.md`。
|
||||
- 自己审阅并完善,确保与你的愿景一致。初期可以很简陋,目标是给 AI 提供游戏结构和意图的上下文。不要过度设计,后续会迭代。
|
||||
</details>
|
||||
|
||||
<details>
|
||||
<summary><strong>2. 技术栈与 <code>CLAUDE.md</code> / <code>Agents.md</code></strong></summary>
|
||||
|
||||
- 让 **gpt-5.1-codex** 或 **Claude Opus 4.5** 为你的游戏推荐最合适的技术栈(例如:多人3D游戏用 ThreeJS + WebSocket),保存为 `tech-stack.md`。
|
||||
- 让 **gpt-5.3-codex** 或 **Claude Opus 4.6** 为你的游戏推荐最合适的技术栈(例如:多人3D游戏用 ThreeJS + WebSocket),保存为 `tech-stack.md`。
|
||||
- 要求它提出 **最简单但最健壮** 的技术栈。
|
||||
- 在终端中打开 **Claude Code** 或 **Codex CLI**,使用 `/init` 命令,它会读取你已创建的两个 .md 文件,生成一套规则来正确引导大模型。
|
||||
- **关键:一定要审查生成的规则。** 确保规则强调 **模块化**(多文件)和禁止 **单体巨文件**(monolith)。可能需要手动修改或补充规则。
|
||||
|
|
@ -413,7 +413,7 @@ gantt
|
|||
<details>
|
||||
<summary><strong>3. 实施计划(Implementation Plan)</strong></summary>
|
||||
|
||||
- 将以下内容提供给 **gpt-5.1-codex** 或 **Claude Opus 4.5**:
|
||||
- 将以下内容提供给 **gpt-5.3-codex** 或 **Claude Opus 4.6**:
|
||||
- 游戏设计文档(`game-design-document.md`)
|
||||
- 技术栈推荐(`tech-stack.md`)
|
||||
- 让它生成一份详细的 **实施计划**(Markdown 格式),包含一系列给 AI 开发者的分步指令。
|
||||
|
|
@ -457,7 +457,7 @@ gantt
|
|||
- 打开 **Codex** 或 **Claude Code**(扩展或终端)。
|
||||
- 提示词:阅读 `/memory-bank` 所有文档,然后执行实施计划的第 1 步。我会负责跑测试。在我验证测试通过前,不要开始第 2 步。验证通过后,打开 `progress.md` 记录你做了什么供后续开发者参考,再把新的架构洞察添加到 `architecture.md` 中解释每个文件的作用。
|
||||
- **永远** 先用 "Ask" 模式或 "Plan Mode"(Claude Code 中按 `shift+tab`),确认满意后再让 AI 执行该步骤。
|
||||
- **极致 Vibe:** 安装 [Superwhisper](https://superwhisper.com),用语音随便跟 Claude 或 gpt-5.1-codex 聊天,不用打字。
|
||||
- **极致 Vibe:** 安装 [Superwhisper](https://superwhisper.com),用语音随便跟 Claude 或 gpt-5.3-codex 聊天,不用打字。
|
||||
</details>
|
||||
|
||||
<details>
|
||||
|
|
@ -489,7 +489,7 @@ gantt
|
|||
<summary><strong>常规修复</strong></summary>
|
||||
|
||||
- 如果某个提示词失败或搞崩了项目:
|
||||
- Claude Code 用 `/rewind` 回退;用 gpt-5.1-codex 的话多提交 git,需要时 reset。
|
||||
- Claude Code 用 `/rewind` 回退;用 gpt-5.3-codex 的话多提交 git,需要时 reset。
|
||||
- 报错处理:
|
||||
- **JavaScript 错误:** 打开浏览器控制台(F12),复制错误,贴给 AI;视觉问题截图发给它。
|
||||
- **懒人方案:** 安装 [BrowserTools](https://browsertools.agentdesk.ai/installation),自动复制错误和截图。
|
||||
|
|
@ -501,7 +501,7 @@ gantt
|
|||
- 实在卡住:
|
||||
- 回退到上一个 git commit(`git reset`),换新提示词重试。
|
||||
- 极度卡壳:
|
||||
- 用 [RepoPrompt](https://repoprompt.com/) 或 [uithub](https://uithub.com/) 把整个代码库合成一个文件,然后丢给 **gpt-5.1-codex 或 Claude** 求救。
|
||||
- 用 [RepoPrompt](https://repoprompt.com/) 或 [uithub](https://uithub.com/) 把整个代码库合成一个文件,然后丢给 **gpt-5.3-codex 或 Claude** 求救。
|
||||
</details>
|
||||
|
||||
</details>
|
||||
|
|
@ -522,7 +522,7 @@ gantt
|
|||
<details>
|
||||
<summary><strong>其他实用技巧</strong></summary>
|
||||
|
||||
- **小修改:** 用 gpt-5.1-codex (medium)
|
||||
- **小修改:** 用 gpt-5.3-codex (medium)
|
||||
- **写顶级营销文案:** 用 Opus 4.1
|
||||
- **生成优秀 2D 精灵图:** 用 ChatGPT + Nano Banana
|
||||
- **生成音乐:** 用 Suno
|
||||
|
|
@ -545,7 +545,7 @@ gantt
|
|||
- **A:** 那不是一个提示词,是 ~30 个提示词 + 专门的 `plane-implementation.md` 文件引导的。用精准指令如“在机翼上为副翼切出空间”,而不是“做一个飞机”这种模糊指令。
|
||||
|
||||
- **Q: 为什么现在 Claude Code 或 Codex CLI 比 Cursor 更强?**
|
||||
- **A:** 完全看个人喜好。我们强调的是:Claude Code 能更好发挥 Claude Opus 4.5 的实力,Codex CLI 能更好发挥 gpt-5.1-codex 的实力,而 Cursor 对这两者的利用都不如原生终端版。终端版还能在任意 IDE、使用 SSH 远程服务器等场景工作,自定义命令、子代理、钩子等功能也能长期大幅提升开发质量和速度。最后,即使你只是低配 Claude 或 ChatGPT 订阅,也完全够用。
|
||||
- **A:** 完全看个人喜好。我们强调的是:Claude Code 能更好发挥 Claude Opus 4.6 的实力,Codex CLI 能更好发挥 gpt-5.3-codex 的实力,而 Cursor 对这两者的利用都不如原生终端版。终端版还能在任意 IDE、使用 SSH 远程服务器等场景工作,自定义命令、子代理、钩子等功能也能长期大幅提升开发质量和速度。最后,即使你只是低配 Claude 或 ChatGPT 订阅,也完全够用。
|
||||
|
||||
- **Q: 我不会搭建多人游戏的服务器怎么办?**
|
||||
- **A:** 问你的 AI。
|
||||
|
|
|
|||
|
|
@ -34,7 +34,7 @@ Required inputs:
|
|||
**Codex CLI**
|
||||
```bash
|
||||
# --yolo is an alias for --dangerously-bypass-approvals-and-sandbox
|
||||
alias c='codex --enable web_search_request -m gpt-5.1-codex-max -c model_reasoning_effort="high" --yolo'
|
||||
alias c='codex --enable web_search_request -m gpt-5.3-codex-max -c model_reasoning_effort="high" --yolo'
|
||||
```
|
||||
|
||||
**Claude Code**
|
||||
|
|
|
|||
|
|
@ -66,8 +66,8 @@ codex login status
|
|||
## Available Models
|
||||
|
||||
- `gpt-5-codex` - Standard model
|
||||
- `gpt-5.1-codex` - Enhanced version
|
||||
- `gpt-5.1-codex-max` - Strongest model
|
||||
- `gpt-5.3-codex` - Enhanced version
|
||||
- `gpt-5.3-codex-max` - Strongest model
|
||||
|
||||
## Reasoning Strength Configuration
|
||||
|
||||
|
|
@ -96,7 +96,7 @@ codex --dangerously-bypass-approvals-and-sandbox "Your prompt"
|
|||
codex --full-auto "Your prompt"
|
||||
|
||||
# Full YOLO config alias
|
||||
alias c='codex --enable web_search_request -m gpt-5.1-codex-max -c model_reasoning_effort="high" --yolo'
|
||||
alias c='codex --enable web_search_request -m gpt-5.3-codex-max -c model_reasoning_effort="high" --yolo'
|
||||
|
||||
# Resume last session
|
||||
codex resume --last
|
||||
|
|
|
|||
|
|
@ -125,7 +125,7 @@ AI CLI 工具默认会询问确认,开启全权限模式可以跳过:
|
|||
|
||||
```bash
|
||||
# Codex - 最强配置
|
||||
codex --enable web_search_request -m gpt-5.1-codex-max -c model_reasoning_effort="high" --dangerously-bypass-approvals-and-sandbox
|
||||
codex --enable web_search_request -m gpt-5.3-codex-max -c model_reasoning_effort="high" --dangerously-bypass-approvals-and-sandbox
|
||||
|
||||
# Claude Code - 跳过所有确认
|
||||
claude --dangerously-skip-permissions
|
||||
|
|
@ -140,7 +140,7 @@ gemini --yolo
|
|||
|
||||
```bash
|
||||
# c - Codex (GPT-5.1 最强模式)
|
||||
alias c='codex --enable web_search_request -m gpt-5.1-codex-max -c model_reasoning_effort="high" --dangerously-bypass-approvals-and-sandbox'
|
||||
alias c='codex --enable web_search_request -m gpt-5.3-codex-max -c model_reasoning_effort="high" --dangerously-bypass-approvals-and-sandbox'
|
||||
|
||||
# cc - Claude Code (全权限)
|
||||
alias cc='claude --dangerously-skip-permissions'
|
||||
|
|
|
|||
|
|
@ -28,7 +28,7 @@ cd /mnt/d/.projects/kiro-account-manager-main.zip/proxycast-main && ./src-tauri/
|
|||
|
||||
## 二、Claude Code 启动命令
|
||||
|
||||
### Claude Opus 4.5 (Kiro 凭证,最强)
|
||||
### Claude Opus 4.6 (Kiro 凭证,最强)
|
||||
```bash
|
||||
CLAUDE_CODE_DISABLE_NONESSENTIAL_TRAFFIC=1 ANTHROPIC_BASE_URL=http://127.0.0.1:8999 ANTHROPIC_API_KEY=proxy_cast claude --dangerously-skip-permissions --model claude-opus-4-5
|
||||
```
|
||||
|
|
@ -146,7 +146,7 @@ curl http://127.0.0.1:8999/v1/models -H "Authorization: Bearer proxy_cast"
|
|||
| `nousresearch/hermes-3-llama-3.1-405b:free` | 指令遵循好 |
|
||||
| `openai/gpt-oss-120b:free` | GPT 开源 120B |
|
||||
| `openai/gpt-oss-20b:free` | GPT 开源 20B |
|
||||
| `moonshotai/kimi-k2:free` | Kimi K2 月之暗面 |
|
||||
| `moonshotai/kimi-k2:free` | Kimi K2.5 月之暗面 |
|
||||
| `z-ai/glm-4.5-air:free` | 中文原生 |
|
||||
| `alibaba/tongyi-deepresearch-30b-a3b:free` | 中文研究 |
|
||||
| `kwaipilot/kat-coder-pro:free` | 代码 Agent |
|
||||
|
|
@ -320,7 +320,7 @@ curl http://127.0.0.1:8999/v1/models -H "Authorization: Bearer proxy_cast"
|
|||
### Kiro 凭证 (Claude 系列)
|
||||
|
||||
```bash
|
||||
# Claude Opus 4.5 - 最强模型
|
||||
# Claude Opus 4.6 - 最强模型
|
||||
CLAUDE_CODE_DISABLE_NONESSENTIAL_TRAFFIC=1 ANTHROPIC_BASE_URL=http://127.0.0.1:8999 ANTHROPIC_API_KEY=proxy_cast claude --dangerously-skip-permissions --model claude-opus-4-5
|
||||
|
||||
# Claude Sonnet 4.5 - 平衡性能
|
||||
|
|
@ -1071,7 +1071,7 @@ CLAUDE_CODE_DISABLE_NONESSENTIAL_TRAFFIC=1 ANTHROPIC_BASE_URL=http://127.0.0.1:8
|
|||
|------|------|------|
|
||||
| `openai/gpt-oss-120b:free` | GPT 开源 120B | ✅ 已测试 |
|
||||
| `openai/gpt-oss-20b:free` | GPT 开源 20B | ✅ 已测试 |
|
||||
| `moonshotai/kimi-k2:free` | Kimi K2 | ✅ 已测试 |
|
||||
| `moonshotai/kimi-k2:free` | Kimi K2.5 | ✅ 已测试 |
|
||||
|
||||
### Claude Code 启动命令
|
||||
|
||||
|
|
@ -1082,7 +1082,7 @@ CLAUDE_CODE_DISABLE_NONESSENTIAL_TRAFFIC=1 ANTHROPIC_BASE_URL=http://127.0.0.1:8
|
|||
# GPT-OSS 20B - 轻量开源模型
|
||||
CLAUDE_CODE_DISABLE_NONESSENTIAL_TRAFFIC=1 ANTHROPIC_BASE_URL=http://127.0.0.1:8999 ANTHROPIC_API_KEY=proxy_cast claude --dangerously-skip-permissions --model openai/gpt-oss-20b:free
|
||||
|
||||
# Kimi K2 - 月之暗面最新模型
|
||||
# Kimi K2.5 - 月之暗面最新模型
|
||||
CLAUDE_CODE_DISABLE_NONESSENTIAL_TRAFFIC=1 ANTHROPIC_BASE_URL=http://127.0.0.1:8999 ANTHROPIC_API_KEY=proxy_cast claude --dangerously-skip-permissions --model moonshotai/kimi-k2:free
|
||||
```
|
||||
|
||||
|
|
@ -1095,7 +1095,7 @@ curl -s http://127.0.0.1:8999/v1/chat/completions \
|
|||
-H "Authorization: Bearer proxy_cast" \
|
||||
-d '{"model":"openai/gpt-oss-120b:free","messages":[{"role":"user","content":"Hi"}],"max_tokens":10}'
|
||||
|
||||
# 测试 Kimi K2
|
||||
# 测试 Kimi K2.5
|
||||
curl -s http://127.0.0.1:8999/v1/chat/completions \
|
||||
-H "Content-Type: application/json" \
|
||||
-H "Authorization: Bearer proxy_cast" \
|
||||
|
|
@ -1111,7 +1111,7 @@ curl -s http://127.0.0.1:8999/v1/chat/completions \
|
|||
| `mistralai/devstral-2512:free` | 代码专精 123B | 编程 |
|
||||
| `openai/gpt-oss-120b:free` | GPT 开源 120B | 通用 |
|
||||
| `openai/gpt-oss-20b:free` | GPT 开源 20B | 快速响应 |
|
||||
| `moonshotai/kimi-k2:free` | Kimi K2 | 中文任务 |
|
||||
| `moonshotai/kimi-k2:free` | Kimi K2.5 | 中文任务 |
|
||||
| `google/gemini-2.0-flash-exp:free` | Gemini 2.0 | 翻译 |
|
||||
| `meta-llama/llama-3.3-70b-instruct:free` | Llama 3.3 | 速度快 |
|
||||
| `nousresearch/hermes-3-llama-3.1-405b:free` | Hermes 3 | 指令遵循 |
|
||||
|
|
|
|||
|
|
@ -71,7 +71,7 @@
|
|||
|
||||
| 平台 | 模型 | 特点 |
|
||||
|:---|:---|:---|
|
||||
| [Claude](https://claude.ai/) | Claude Opus 4.5 | 代码能力强,支持 Artifacts |
|
||||
| [Claude](https://claude.ai/) | Claude Opus 4.6 | 代码能力强,支持 Artifacts |
|
||||
| [ChatGPT](https://chatgpt.com/) | GPT-5.1 | 综合能力强,支持 Codex |
|
||||
| [Gemini](https://gemini.google.com/) | Gemini 3.0 Pro | 免费额度大,支持长上下文 |
|
||||
|
||||
|
|
@ -79,7 +79,7 @@
|
|||
|
||||
| 平台 | 模型 | 特点 |
|
||||
|:---|:---|:---|
|
||||
| [Kimi](https://kimi.moonshot.cn/) | Kimi K2 | 长文本处理强 |
|
||||
| [Kimi](https://kimi.moonshot.cn/) | Kimi K2.5 | 长文本处理强 |
|
||||
| [通义千问](https://tongyi.aliyun.com/) | Qwen | 阿里出品,免费 |
|
||||
| [智谱清言](https://chatglm.cn/) | GLM-4 | 智谱 AI 出品 |
|
||||
| [豆包](https://www.doubao.com/) | 豆包 | 字节出品 |
|
||||
|
|
@ -129,7 +129,7 @@
|
|||
|
||||
| 资源 | 链接 | 说明 |
|
||||
|:---|:---|:---|
|
||||
| Kiro | [kiro.dev](https://kiro.dev/) | 免费 Claude Opus 4.5 |
|
||||
| Kiro | [kiro.dev](https://kiro.dev/) | 免费 Claude Opus 4.6 |
|
||||
| Windsurf | [windsurf.com](https://windsurf.com/) | 新用户免费额度 |
|
||||
| GitHub Copilot | [github.com/copilot](https://github.com/copilot) | 学生/开源免费 |
|
||||
| Codeium | [codeium.com](https://codeium.com/) | 免费 AI 代码补全 |
|
||||
|
|
|
|||
|
|
@ -17,7 +17,7 @@
|
|||
|
||||
| 模型 | 说明 |
|
||||
|:---|:---|
|
||||
| Claude Opus 4.5 | 代码能力强 |
|
||||
| Claude Opus 4.6 | 代码能力强 |
|
||||
| GPT-5.1 Codex | 复杂逻辑处理 |
|
||||
| Gemini 2.5 Pro | 免费长上下文 |
|
||||
|
||||
|
|
|
|||
|
|
@ -83,12 +83,12 @@
|
|||
|
||||
## 📋 器
|
||||
|
||||
- [**Claude Opus 4.5**](https://claude.ai/new),在 Claude Code 中使用 很贵,但是尼区ios订阅要便宜几百人民币,快+效果好,顶中顶中顶,有 cli 和 ide 插件
|
||||
- [**gpt-5.1-codex.1-codex (xhigh)**](https://chatgpt.com/codex/),在 Codex CLI 中使用,顶中顶,除了慢其他没得挑,大项目复杂逻辑唯一解,买chatgpt会员就能用,有 cli 和 ide 插件
|
||||
- [**Droid**](https://factory.ai/news/terminal-bench),这个里面的 Claude Opus 4.5比 Claude Code 还强,顶,有 cli
|
||||
- [**Kiro**](https://kiro.dev/),这个里面的 Claude Opus 4.5 现在免费,就是cli有点拉,看不到正在运行的情况有客户端和 cli
|
||||
- [**Claude Opus 4.6**](https://claude.ai/new),在 Claude Code 中使用 很贵,但是尼区ios订阅要便宜几百人民币,快+效果好,顶中顶中顶,有 cli 和 ide 插件
|
||||
- [**gpt-5.3-codex (xhigh)**](https://chatgpt.com/codex/),在 Codex CLI 中使用,顶中顶,除了慢其他没得挑,大项目复杂逻辑唯一解,买chatgpt会员就能用,有 cli 和 ide 插件
|
||||
- [**Droid**](https://factory.ai/news/terminal-bench),这个里面的 Claude Opus 4.6比 Claude Code 还强,顶,有 cli
|
||||
- [**Kiro**](https://kiro.dev/),这个里面的 Claude Opus 4.6 现在免费,就是cli有点拉,看不到正在运行的情况有客户端和 cli
|
||||
- [**gemini**](https://geminicli.com/),目前免费用,干脏活,用 Claude Code 或者 codex 写好的脚本,拿他来执行可以,整理文档和找思路就它了有客户端和 cli
|
||||
- [**antigravity**](https://antigravity.google/),谷歌的,可以免费用 Claude Opus 4.5 和 gemini 3.0 pro 大善人
|
||||
- [**antigravity**](https://antigravity.google/),谷歌的,可以免费用 Claude Opus 4.6 和 gemini 3.0 pro 大善人
|
||||
- [**aistudio**](https://aistudio.google.com/prompts/new_chat),谷歌家的,免费用 gemini 3.0 pro 和 Nano Banana
|
||||
- [**gemini-enterprise**](https://cloud.google.com/gemini-enterprise),谷歌企业版,现在能免费用 Nano Banana pro
|
||||
- [**augment**](https://app.augmentcode.com/),它的上下文引擎和提示词优化按钮真的神中神中神,小白就用它就行了,点击按钮自动帮你写好提示词,懒人必备
|
||||
|
|
@ -370,12 +370,12 @@ gantt
|
|||
|
||||
## 🚀 入门指南(这里是原作者的,不是我写的,我更新了一下我认为最好的模型)
|
||||
要开始 Vibe Coding,你只需要以下两种工具之一:
|
||||
- **Claude Opus 4.5**,在 Claude Code 中使用
|
||||
- **gpt-5.1-codex.1-codex (xhigh)**,在 Codex CLI 中使用
|
||||
- **Claude Opus 4.6**,在 Claude Code 中使用
|
||||
- **gpt-5.3-codex (xhigh)**,在 Codex CLI 中使用
|
||||
|
||||
本指南同时适用于 CLI 终端版本和 VSCode 扩展版本(Codex 和 Claude Code 都有扩展,且界面更新)。
|
||||
|
||||
*(注:本指南早期版本使用的是 **Grok 3**,后来切换到 **Gemini 2.5 Pro**,现在我们使用的是 **Claude 4.5**(或 **gpt-5.1-codex.1-codex (xhigh)**))*
|
||||
*(注:本指南早期版本使用的是 **Grok 3**,后来切换到 **Gemini 2.5 Pro**,现在我们使用的是 **Claude 4.5**(或 **gpt-5.3-codex (xhigh)**))*
|
||||
|
||||
*(注2:如果你想使用 Cursor,请查看本指南的 [1.1 版本](https://github.com/EnzeD/vibe-coding/tree/1.1.1),但我们认为它目前不如 Codex CLI 或 Claude Code 强大)*
|
||||
|
||||
|
|
@ -387,14 +387,14 @@ gantt
|
|||
<details>
|
||||
<summary><strong>1. 游戏设计文档(Game Design Document)</strong></summary>
|
||||
|
||||
- 把你的游戏创意交给 **gpt-5.1-codex** 或 **Claude Opus 4.5**,让它生成一份简洁的 **游戏设计文档**,格式为 Markdown,文件名为 `game-design-document.md`。
|
||||
- 把你的游戏创意交给 **gpt-5.3-codex** 或 **Claude Opus 4.6**,让它生成一份简洁的 **游戏设计文档**,格式为 Markdown,文件名为 `game-design-document.md`。
|
||||
- 自己审阅并完善,确保与你的愿景一致。初期可以很简陋,目标是给 AI 提供游戏结构和意图的上下文。不要过度设计,后续会迭代。
|
||||
</details>
|
||||
|
||||
<details>
|
||||
<summary><strong>2. 技术栈与 <code>CLAUDE.md</code> / <code>Agents.md</code></strong></summary>
|
||||
|
||||
- 让 **gpt-5.1-codex** 或 **Claude Opus 4.5** 为你的游戏推荐最合适的技术栈(例如:多人3D游戏用 ThreeJS + WebSocket),保存为 `tech-stack.md`。
|
||||
- 让 **gpt-5.3-codex** 或 **Claude Opus 4.6** 为你的游戏推荐最合适的技术栈(例如:多人3D游戏用 ThreeJS + WebSocket),保存为 `tech-stack.md`。
|
||||
- 要求它提出 **最简单但最健壮** 的技术栈。
|
||||
- 在终端中打开 **Claude Code** 或 **Codex CLI**,使用 `/init` 命令,它会读取你已创建的两个 .md 文件,生成一套规则来正确引导大模型。
|
||||
- **关键:一定要审查生成的规则。** 确保规则强调 **模块化**(多文件)和禁止 **单体巨文件**(monolith)。可能需要手动修改或补充规则。
|
||||
|
|
@ -412,7 +412,7 @@ gantt
|
|||
<details>
|
||||
<summary><strong>3. 实施计划(Implementation Plan)</strong></summary>
|
||||
|
||||
- 将以下内容提供给 **gpt-5.1-codex** 或 **Claude Opus 4.5**:
|
||||
- 将以下内容提供给 **gpt-5.3-codex** 或 **Claude Opus 4.6**:
|
||||
- 游戏设计文档(`game-design-document.md`)
|
||||
- 技术栈推荐(`tech-stack.md`)
|
||||
- 让它生成一份详细的 **实施计划**(Markdown 格式),包含一系列给 AI 开发者的分步指令。
|
||||
|
|
@ -456,7 +456,7 @@ gantt
|
|||
- 打开 **Codex** 或 **Claude Code**(扩展或终端)。
|
||||
- 提示词:阅读 `/memory-bank` 所有文档,然后执行实施计划的第 1 步。我会负责跑测试。在我验证测试通过前,不要开始第 2 步。验证通过后,打开 `progress.md` 记录你做了什么供后续开发者参考,再把新的架构洞察添加到 `architecture.md` 中解释每个文件的作用。
|
||||
- **永远** 先用 "Ask" 模式或 "Plan Mode"(Claude Code 中按 `shift+tab`),确认满意后再让 AI 执行该步骤。
|
||||
- **极致 Vibe:** 安装 [Superwhisper](https://superwhisper.com),用语音随便跟 Claude 或 gpt-5.1-codex 聊天,不用打字。
|
||||
- **极致 Vibe:** 安装 [Superwhisper](https://superwhisper.com),用语音随便跟 Claude 或 gpt-5.3-codex 聊天,不用打字。
|
||||
</details>
|
||||
|
||||
<details>
|
||||
|
|
@ -488,7 +488,7 @@ gantt
|
|||
<summary><strong>常规修复</strong></summary>
|
||||
|
||||
- 如果某个提示词失败或搞崩了项目:
|
||||
- Claude Code 用 `/rewind` 回退;用 gpt-5.1-codex 的话多提交 git,需要时 reset。
|
||||
- Claude Code 用 `/rewind` 回退;用 gpt-5.3-codex 的话多提交 git,需要时 reset。
|
||||
- 报错处理:
|
||||
- **JavaScript 错误:** 打开浏览器控制台(F12),复制错误,贴给 AI;视觉问题截图发给它。
|
||||
- **懒人方案:** 安装 [BrowserTools](https://browsertools.agentdesk.ai/installation),自动复制错误和截图。
|
||||
|
|
@ -500,7 +500,7 @@ gantt
|
|||
- 实在卡住:
|
||||
- 回退到上一个 git commit(`git reset`),换新提示词重试。
|
||||
- 极度卡壳:
|
||||
- 用 [RepoPrompt](https://repoprompt.com/) 或 [uithub](https://uithub.com/) 把整个代码库合成一个文件,然后丢给 **gpt-5.1-codex 或 Claude** 求救。
|
||||
- 用 [RepoPrompt](https://repoprompt.com/) 或 [uithub](https://uithub.com/) 把整个代码库合成一个文件,然后丢给 **gpt-5.3-codex 或 Claude** 求救。
|
||||
</details>
|
||||
|
||||
</details>
|
||||
|
|
@ -521,7 +521,7 @@ gantt
|
|||
<details>
|
||||
<summary><strong>其他实用技巧</strong></summary>
|
||||
|
||||
- **小修改:** 用 gpt-5.1-codex (medium)
|
||||
- **小修改:** 用 gpt-5.3-codex (medium)
|
||||
- **写顶级营销文案:** 用 Opus 4.1
|
||||
- **生成优秀 2D 精灵图:** 用 ChatGPT + Nano Banana
|
||||
- **生成音乐:** 用 Suno
|
||||
|
|
@ -544,7 +544,7 @@ gantt
|
|||
- **A:** 那不是一个提示词,是 ~30 个提示词 + 专门的 `plane-implementation.md` 文件引导的。用精准指令如“在机翼上为副翼切出空间”,而不是“做一个飞机”这种模糊指令。
|
||||
|
||||
- **Q: 为什么现在 Claude Code 或 Codex CLI 比 Cursor 更强?**
|
||||
- **A:** 完全看个人喜好。我们强调的是:Claude Code 能更好发挥 Claude Opus 4.5 的实力,Codex CLI 能更好发挥 gpt-5.1-codex 的实力,而 Cursor 对这两者的利用都不如原生终端版。终端版还能在任意 IDE、使用 SSH 远程服务器等场景工作,自定义命令、子代理、钩子等功能也能长期大幅提升开发质量和速度。最后,即使你只是低配 Claude 或 ChatGPT 订阅,也完全够用。
|
||||
- **A:** 完全看个人喜好。我们强调的是:Claude Code 能更好发挥 Claude Opus 4.6 的实力,Codex CLI 能更好发挥 gpt-5.3-codex 的实力,而 Cursor 对这两者的利用都不如原生终端版。终端版还能在任意 IDE、使用 SSH 远程服务器等场景工作,自定义命令、子代理、钩子等功能也能长期大幅提升开发质量和速度。最后,即使你只是低配 Claude 或 ChatGPT 订阅,也完全够用。
|
||||
|
||||
- **Q: 我不会搭建多人游戏的服务器怎么办?**
|
||||
- **A:** 问你的 AI。
|
||||
|
|
|
|||
|
|
@ -34,7 +34,7 @@ description: "无头模式 AI CLI 调用技能:支持 Gemini/Claude/Codex CLI
|
|||
**Codex CLI**
|
||||
```bash
|
||||
# --yolo 是 --dangerously-bypass-approvals-and-sandbox 的别名
|
||||
alias c='codex --enable web_search_request -m gpt-5.1-codex-max -c model_reasoning_effort="high" --yolo'
|
||||
alias c='codex --enable web_search_request -m gpt-5.3-codex-max -c model_reasoning_effort="high" --yolo'
|
||||
```
|
||||
|
||||
**Claude Code**
|
||||
|
|
|
|||
|
|
@ -65,8 +65,8 @@ codex login status
|
|||
## 可用模型
|
||||
|
||||
- `gpt-5-codex` - 标准模型
|
||||
- `gpt-5.1-codex` - 增强版
|
||||
- `gpt-5.1-codex-max` - 最强模型
|
||||
- `gpt-5.3-codex` - 增强版
|
||||
- `gpt-5.3-codex-max` - 最强模型
|
||||
|
||||
## 推理强度配置
|
||||
|
||||
|
|
@ -95,7 +95,7 @@ codex --dangerously-bypass-approvals-and-sandbox "Your prompt"
|
|||
codex --full-auto "Your prompt"
|
||||
|
||||
# 完整 YOLO 配置别名
|
||||
alias c='codex --enable web_search_request -m gpt-5.1-codex-max -c model_reasoning_effort="high" --yolo'
|
||||
alias c='codex --enable web_search_request -m gpt-5.3-codex-max -c model_reasoning_effort="high" --yolo'
|
||||
|
||||
# 恢复最近会话
|
||||
codex resume --last
|
||||
|
|
|
|||
Loading…
Reference in New Issue