mirror of
https://github.com/GuDaStudio/skills.git
synced 2026-02-04 18:40:41 +08:00
v0.1.0：发布测试版
This commit is contained in:
GuDaStudio
72
.github/ISSUE_TEMPLATE/bug_report.md
vendored
Normal file
72
.github/ISSUE_TEMPLATE/bug_report.md
vendored
Normal file
@@ -0,0 +1,72 @@
|
|||||||
|
---
|
||||||
|
name: Bug report
|
||||||
|
about: Report a bug to help us improve GudaCC Skills
|
||||||
|
title: "[BUG] "
|
||||||
|
labels: bug
|
||||||
|
assignees: GuDaStudio
|
||||||
|
|
||||||
|
---
|
||||||
|
|
||||||
|
## Bug Description / Bug描述
|
||||||
|
A clear and concise description of the bug.
|
||||||
|
清晰简洁的错误描述。
|
||||||
|
|
||||||
|
## Environment / 运行环境
|
||||||
|
- **OS**: e.g., Ubuntu 22.04 / macOS 14.0 / Windows 11 / WSL2 ...
|
||||||
|
- **Skills Installation**: e.g., user-level (`~/.claude/skills/`) / project-level (`./.claude/skills/`)
|
||||||
|
- **Claude Code Version**: run `claude --version`
|
||||||
|
- **Codex CLI Version** (if using collaborating-with-codex): run `codex --version`
|
||||||
|
- **Gemini CLI Version** (if using collaborating-with-gemini): run `gemini --version`
|
||||||
|
|
||||||
|
## Affected Skill / 受影响的 Skill
|
||||||
|
- [ ] collaborating-with-codex
|
||||||
|
- [ ] collaborating-with-gemini
|
||||||
|
- [ ] Other / 其他
|
||||||
|
|
||||||
|
## Steps to Reproduce / 复现步骤
|
||||||
|
1.
|
||||||
|
2.
|
||||||
|
3.
|
||||||
|
|
||||||
|
## Logs / 错误日志
|
||||||
|
If the `error` field is returned, please paste it here.
|
||||||
|
如有`error`字段返回，请粘贴在这里。
|
||||||
|
```json
|
||||||
|
|
||||||
|
```
|
||||||
|
|
||||||
|
## Skill Call Parameters / Skill 调用参数
|
||||||
|
If the bug is related to calling a skill, please provide the parameters used.
|
||||||
|
若该故障与调用 Skill 相关，请提供所用参数。
|
||||||
|
|
||||||
|
<details>
|
||||||
|
<summary>collaborating-with-codex</summary>
|
||||||
|
|
||||||
|
```json
|
||||||
|
{
|
||||||
|
"PROMPT": "",
|
||||||
|
"cd": "",
|
||||||
|
"sandbox": "read-only",
|
||||||
|
"SESSION_ID": "",
|
||||||
|
"return_all_messages": false
|
||||||
|
}
|
||||||
|
```
|
||||||
|
</details>
|
||||||
|
|
||||||
|
<details>
|
||||||
|
<summary>collaborating-with-gemini</summary>
|
||||||
|
|
||||||
|
```json
|
||||||
|
{
|
||||||
|
"PROMPT": "",
|
||||||
|
"cd": "",
|
||||||
|
"sandbox": false,
|
||||||
|
"SESSION_ID": "",
|
||||||
|
"return_all_messages": false
|
||||||
|
}
|
||||||
|
```
|
||||||
|
</details>
|
||||||
|
|
||||||
|
## Additional Context / 其他信息
|
||||||
|
Additional background information regarding this issue, such as screenshots or related discussions. We highly recommend including complete report screenshots!
|
||||||
|
关于该问题的其他背景信息，例如截图或相关议题。我们十分建议您添加完整的报告截图！
|
||||||
55
.github/ISSUE_TEMPLATE/feature-request.md
vendored
Normal file
55
.github/ISSUE_TEMPLATE/feature-request.md
vendored
Normal file
@@ -0,0 +1,55 @@
|
|||||||
|
---
|
||||||
|
name: Feature Request
|
||||||
|
about: Suggest an idea or enhancement for GudaCC Skills
|
||||||
|
title: "[Feature] "
|
||||||
|
labels: enhancement
|
||||||
|
assignees: GuDaStudio
|
||||||
|
|
||||||
|
---
|
||||||
|
|
||||||
|
## Feature Summary / 功能概述
|
||||||
|
A clear and concise description of the feature you'd like to see.
|
||||||
|
清晰简洁地描述您希望看到的功能。
|
||||||
|
|
||||||
|
## Target Skill / 目标 Skill
|
||||||
|
- [ ] collaborating-with-codex
|
||||||
|
- [ ] collaborating-with-gemini
|
||||||
|
- [ ] New Skill / 新 Skill
|
||||||
|
- [ ] General / 通用
|
||||||
|
|
||||||
|
## Problem / Motivation / 问题背景
|
||||||
|
Describe the problem or limitation that this feature would address.
|
||||||
|
描述此功能将解决的问题或限制。
|
||||||
|
|
||||||
|
**Is your feature request related to a problem?**
|
||||||
|
**您的功能请求是否与某个问题相关？**
|
||||||
|
|
||||||
|
Example: "I'm always frustrated when [...]"
|
||||||
|
例如："当 [...] 时，我总是感到困扰"
|
||||||
|
|
||||||
|
## Proposed Solution / 建议方案
|
||||||
|
Describe how you envision the feature working.
|
||||||
|
描述您设想的功能运作方式。
|
||||||
|
|
||||||
|
## Alternatives Considered / 其他方案
|
||||||
|
Describe any alternative solutions or features you've considered.
|
||||||
|
描述您考虑过的任何替代方案或功能。
|
||||||
|
|
||||||
|
## Use Case / 使用场景
|
||||||
|
Provide examples of how this feature would be used.
|
||||||
|
提供此功能的使用示例。
|
||||||
|
|
||||||
|
```
|
||||||
|
Example workflow or code snippet here
|
||||||
|
示例工作流程或代码片段
|
||||||
|
```
|
||||||
|
|
||||||
|
## Environment (if relevant) / 运行环境（如相关）
|
||||||
|
- **OS**: e.g., Ubuntu 22.04 / macOS 14.0 / Windows 11 / WSL2 ...
|
||||||
|
- **Claude Code Version**: run `claude --version`
|
||||||
|
- **Codex CLI Version** (if relevant): run `codex --version`
|
||||||
|
- **Gemini CLI Version** (if relevant): run `gemini --version`
|
||||||
|
|
||||||
|
## Additional Context / 其他信息
|
||||||
|
Any other context, mockups, or screenshots about the feature request.
|
||||||
|
关于此功能请求的其他背景信息、原型图或截图。
|
||||||
368
README.md
Normal file
368
README.md
Normal file
@@ -0,0 +1,368 @@
|
|||||||
|
|
||||||
|
|
||||||
|
<div align="center">
|
||||||
|
|
||||||
|
**Agent Skills 集合 —— 让 Claude 与多模型/工具无缝协作**
|
||||||
|
|
||||||
|
[](https://opensource.org/licenses/MIT) [](https://claude.ai/code) [](https://x.com/intent/tweet?text=GudaCC%20Skills：Claude%20Code%20Agent%20Skills%20集合%20https://github.com/GuDaStudio/skills%20%23AI%20%23Coding%20%23ClaudeCode) [](https://www.facebook.com/sharer/sharer.php?u=https://github.com/GuDaStudio/skills) [](https://t.me/share/url?url=https://github.com/GuDaStudio/skills&text=GudaCC%20Skills：Claude%20Code%20Agent%20Skills%20集合)
|
||||||
|
|
||||||
|
⭐ 在 GitHub 上给我们点星~您的支持对我们意义重大！ 🙏😊
|
||||||
|
|
||||||
|
[English](./docs/README_EN.md) | 简体中文
|
||||||
|
|
||||||
|
</div>
|
||||||
|
|
||||||
|
---
|
||||||
|
|
||||||
|
## 一、项目简介
|
||||||
|
|
||||||
|
**Agent Skills** 是 [Anthropic 推出的](https://platform.claude.com/docs/en/agents-and-tools/agent-skills/overview)模块化能力扩展机制，让 LLM 能够按需加载专业领域知识与工作流。本仓库收录了 **GudaStudio** 开发的 Agent Skills 集合，实现 Claude 与其他 AI 模型、工具的无缝协作。
|
||||||
|
|
||||||
|
### 已收录的 Skills
|
||||||
|
|
||||||
|
| Skill | 功能描述 | 协作模型 |
|
||||||
|
|-------|----------|----------|
|
||||||
|
| [collaborating-with-codex](./collaborating-with-codex) | 将编码任务委托给 Codex CLI，用于原型开发、调试和代码审查 | OpenAI Codex |
|
||||||
|
| [collaborating-with-gemini](./collaborating-with-gemini) | 将编码任务委托给 Gemini CLI，用于原型开发、调试和代码审查 | Google Gemini |
|
||||||
|
|
||||||
|
---
|
||||||
|
|
||||||
|
## 二、快速开始
|
||||||
|
|
||||||
|
### 0. 前置要求
|
||||||
|
|
||||||
|
请确保已安装并配置以下工具：
|
||||||
|
|
||||||
|
- [Claude Code](https://docs.claude.com/docs/claude-code) (v2.0.56+)
|
||||||
|
- [Codex CLI](https://github.com/openai/codex) (若使用 collaborating-with-codex)
|
||||||
|
- [Gemini CLI](https://github.com/google-gemini/gemini-cli) (若使用 collaborating-with-gemini)
|
||||||
|
|
||||||
|
### 1. 获取仓库
|
||||||
|
|
||||||
|
```bash
|
||||||
|
# 克隆到任意位置
|
||||||
|
git clone --recurse-submodules https://github.com/GuDaStudio/skills
|
||||||
|
cd skills
|
||||||
|
```
|
||||||
|
|
||||||
|
### 2. 安装 Skills
|
||||||
|
|
||||||
|
提供安装脚本，支持灵活选择安装范围和目标位置。
|
||||||
|
|
||||||
|
**查看可用 Skills：**
|
||||||
|
|
||||||
|
```bash
|
||||||
|
./install.sh --list
|
||||||
|
```
|
||||||
|
|
||||||
|
**方式一：一键安装所有 Skills**
|
||||||
|
|
||||||
|
```bash
|
||||||
|
# 用户级安装（所有项目生效）
|
||||||
|
./install.sh --user --all
|
||||||
|
|
||||||
|
# 项目级安装（仅当前项目生效，需在项目根目录执行）
|
||||||
|
./install.sh --project --all
|
||||||
|
```
|
||||||
|
|
||||||
|
**方式二：选择性安装**
|
||||||
|
|
||||||
|
```bash
|
||||||
|
# 仅安装 collaborating-with-codex
|
||||||
|
./install.sh --user --skill collaborating-with-codex
|
||||||
|
|
||||||
|
# 安装多个指定 Skill
|
||||||
|
./install.sh --user -s collaborating-with-codex -s collaborating-with-gemini
|
||||||
|
```
|
||||||
|
|
||||||
|
**方式三：自定义安装路径**
|
||||||
|
|
||||||
|
```bash
|
||||||
|
./install.sh --target /your/custom/path --all
|
||||||
|
```
|
||||||
|
|
||||||
|
<details>
|
||||||
|
<summary>点击查看完整参数说明</summary>
|
||||||
|
|
||||||
|
| 参数 | 简写 | 说明 |
|
||||||
|
|------|------|------|
|
||||||
|
| `--user` | `-u` | 安装到用户级目录 (`~/.claude/skills/`) |
|
||||||
|
| `--project` | `-p` | 安装到项目级目录 (`./.claude/skills/`) |
|
||||||
|
| `--target <path>` | `-t` | 安装到自定义路径 |
|
||||||
|
| `--all` | `-a` | 安装所有可用 Skills |
|
||||||
|
| `--skill <name>` | `-s` | 安装指定 Skill（可多次使用） |
|
||||||
|
| `--list` | `-l` | 列出所有可用 Skills |
|
||||||
|
| `--help` | `-h` | 显示帮助信息 |
|
||||||
|
|
||||||
|
</details>
|
||||||
|
|
||||||
|
### 3. 验证安装
|
||||||
|
|
||||||
|
启动 Claude Code 后，Skills 会自动加载。可通过以下方式验证：
|
||||||
|
|
||||||
|
- 让claude列出所有skill：`list all available SKILLs please`
|
||||||
|
- Claude 应能识别并调用 `collaborating-with-codex` 和 `collaborating-with-gemini`
|
||||||
|
|
||||||
|
|
||||||
|
### 4. （🎊可选，但强烈推荐）配置提示词
|
||||||
|
为让本SKILLS集合更加Claude Code等CLI，我们强烈推荐您在 `~/.claude/CLAUDE.md`中 配置/追加 以下提示词。
|
||||||
|
|
||||||
|
````markdow
|
||||||
|
# CLAUDE.md
|
||||||
|
|
||||||
|
## 0. Global Protocols
|
||||||
|
|
||||||
|
所有操作必须严格遵循以下系统约束：
|
||||||
|
|
||||||
|
- **交互语言**：与工具或模型交互强制使用 **English**；用户输出强制使用 **中文**。
|
||||||
|
- **多轮对话**：如果工具或模型返回的有可持续对话字段 ，比如 `SESSION_ID`，表明工具支持多轮对话，此时记录该字段，并在随后的工具调用中**强制思考**，是否继续进行对话。例如， Codex/Gemini有时会因内部工具调用中断会话，若没有得到需要的回复，则应继续对话。
|
||||||
|
- **沙箱安全**：严禁 Codex/Gemini 对文件系统进行写操作。所有代码获取必须在PROMPT中**明确要求**， 返回 `Unified Diff Patch`，严禁Codex/Gemini做任何真实修改。
|
||||||
|
- **代码主权**：外部模型生成的代码仅作为逻辑参考（Prototype），最终交付代码**必须经过重构**，确保无冗余、企业级标准。
|
||||||
|
- **风格定义**：整体代码风格**始终定位**为，精简高效、毫无冗余。该要求同样适用于注释与文档，且对于这两者，严格遵循**非必要不形成**的核心原则。
|
||||||
|
- **仅对需求做针对性改动**，严禁影响用户现有的其他功能。
|
||||||
|
- Codex/Gemini 互动方式以SKILL形式给出，**强制积极查看、调用**。两者均需要较长调用时长（系统默认"BASH_DEFAULT_TIMEOUT_MS": "300000"），耐心等待即可。
|
||||||
|
- 当检测到可并行化执行的任务时，尽一切可能以并行方式执行。例如当有多个 SKILL 相关的bash命令存在时，可使用`run in background`方式**立即**对需要长时间运行的程序进行挂起以便执行下一个命令，达到并行目的。
|
||||||
|
- 强制遵循 1. Workflow 中的所有Phase，严禁遗漏任何过程。
|
||||||
|
|
||||||
|
## 1. Workflow
|
||||||
|
|
||||||
|
### Phase 1: 上下文全量检索 (Auggie Interface)
|
||||||
|
|
||||||
|
**执行条件**：在生成任何建议或代码前。
|
||||||
|
|
||||||
|
1. **工具调用**：调用 `mcp__auggie-mcp__codebase-retrieval`。
|
||||||
|
2. **检索策略**：
|
||||||
|
- 禁止基于假设（Assumption）回答。
|
||||||
|
- 使用自然语言（NL）构建语义查询（Where/What/How）。
|
||||||
|
- **完整性检查**：必须获取相关类、函数、变量的完整定义与签名。若上下文不足，触发递归检索。
|
||||||
|
3. **需求对齐**：若检索后需求仍有模糊空间，**必须**向用户输出引导性问题列表，直至需求边界清晰（无遗漏、无冗余）。
|
||||||
|
|
||||||
|
### Phase 2: 多模型协作分析 (Analysis & Strategy)
|
||||||
|
|
||||||
|
**执行条件**：上下文就绪后，编码开始前。
|
||||||
|
|
||||||
|
1. **分发输入**：：将用户的**原始需求**（不带预设观点）分发给 Codex 和 Gemini。注意，Codex/Gemini都有完善的CLI系统，所以**无需给出过多上下文**。
|
||||||
|
2. **方案迭代**：
|
||||||
|
- 要求模型提供多角度解决方案。
|
||||||
|
- 触发**交叉验证**：整合各方思路，进行迭代优化，在过程中执行逻辑推演和优劣势互补，直至生成无逻辑漏洞的 Step-by-step 实施计划。
|
||||||
|
3. **用户确认**：向用户展示最终实施计划（含适度伪代码）。
|
||||||
|
|
||||||
|
### Phase 3: 原型获取 (Prototyping)
|
||||||
|
|
||||||
|
**执行条件**：实施计划确认后。根据任务类型路由：
|
||||||
|
|
||||||
|
- **Route A: 前端/UI/样式 (Gemini Kernel)**
|
||||||
|
- **限制**：上下文 < 32k。gemini对于后端逻辑的理解有缺陷，其回复需要客观审视。
|
||||||
|
- **指令**：请求 CSS/React/Vue 原型。以此为最终前端设计原型与视觉基准。
|
||||||
|
- **Route B: 后端/逻辑/算法 (Codex Kernel)**
|
||||||
|
- **能力**：利用其逻辑运算与 Debug 能力。
|
||||||
|
- **指令**：请求逻辑实现原型。
|
||||||
|
- **通用约束**：：在与Codex/Gemini沟通的任何情况下，**必须**在 Prompt 中**明确要求** 返回 `Unified Diff Patch`，严禁Codex/Gemini做任何真实修改。
|
||||||
|
|
||||||
|
### Phase 4: 编码实施 (Implementation)
|
||||||
|
|
||||||
|
**执行准则**：
|
||||||
|
|
||||||
|
1. **逻辑重构**：基于 Phase 3 的原型，去除冗余，**重写**为高可读、高可维护性、企业发布级代码。
|
||||||
|
2. **文档规范**：非必要不生成注释与文档，代码自解释。
|
||||||
|
3. **最小作用域**：变更仅限需求范围，**强制审查**变更是否引入副作用并做针对性修正。
|
||||||
|
|
||||||
|
### Phase 5: 审计与交付 (Audit & Delivery)
|
||||||
|
|
||||||
|
1. **自动审计**：变更生效后，**强制立即调用** Codex与Gemini 同时进行 Code Review，并进行整合修复。
|
||||||
|
- 检查项：逻辑正确性、需求覆盖率、潜在 Bug。
|
||||||
|
2. **交付**：审计通过后反馈给用户。
|
||||||
|
|
||||||
|
## 2. Resource Matrix
|
||||||
|
|
||||||
|
此矩阵定义了各阶段的**强制性**资源调用策略。Claude 作为**主控模型 (Orchestrator)**，必须严格根据当前 Workflow 阶段，按以下规格调度外部资源。
|
||||||
|
|
||||||
|
| Workflow Phase | Functionality | Designated Model / Tool | Input Strategy (Prompting) | Strict Output Constraints | Critical Constraints & Behavior |
|
||||||
|
| :----------------------- | :---------------------- | :--------------------------------------- | :----------------------------------------------------------- | :-------------------------------------------------- | :----------------------------------------------------------- |
|
||||||
|
| **Phase 1** | **Context Retrieval** | **Auggie** (`mcp__auggie`) | **Natural Language (English)**<br>Focus on: *What, Where, How* | **Raw Code / Definitions**<br>(Complete Signatures) | • **Forbidden:** `grep` / keyword search.<br>• **Mandatory:** Recursive retrieval until context is complete. |
|
||||||
|
| **Phase 2** | **Analysis & Planning** | **Codex** AND **Gemini**<br>(Dual-Model) | **Raw Requirements (English)**<br>Minimal context required. | **Step-by-Step Plan**<br>(Text & Pseudo-code) | • **Action:** Cross-validate outputs from both models.<br>• **Goal:** Eliminate logic gaps before coding starts. |
|
||||||
|
| **Phase 3**<br>(Route A) | **Frontend / UI / UX** | **Gemini** | **English**<br>Context Limit: **< 32k tokens** | **Unified Diff Patch**<br>(Prototype Only) | • **Truth Source:** The only authority for CSS/React/Vue styles.<br>• **Warning:** Ignore its backend logic suggestions. |
|
||||||
|
| **Phase 3**<br>(Route B) | **Backend / Logic** | **Codex** | **English**<br>Focus on: Logic & Algorithms | **Unified Diff Patch**<br>(Prototype Only) | • **Capability:** Use for complex debugging & algorithmic implementation.<br>• **Security:** **NO** file system write access allowed. |
|
||||||
|
| **Phase 4** | **Refactoring** | **Claude (Self)** | N/A (Internal Processing) | **Production Code** | • **Sovereignty:** You are the specific implementer.<br>• **Style:** Clean, efficient, no redundancy. Minimal comments. |
|
||||||
|
| **Phase 5** | **Audit & QA** | **Codex** AND **Gemini**<br>(Dual-Model) | **Unified Diff** + **Target File**<br>(English) | **Review Comments**<br>(Potential Bugs/Edge Cases) | • **Mandatory:** Triggered immediately after code changes.<br>• **Action:** Synthesize feedback into a final fix. |
|
||||||
|
````
|
||||||
|
|
||||||
|
---
|
||||||
|
|
||||||
|
## 三、Skills 详解
|
||||||
|
|
||||||
|
### collaborating-with-codex
|
||||||
|
|
||||||
|
将编码任务委托给 OpenAI Codex CLI，适用于算法实现、Bug 分析和代码审查。
|
||||||
|
|
||||||
|
<details>
|
||||||
|
<summary>点击查看参数说明</summary>
|
||||||
|
|
||||||
|
| 参数 | 类型 | 必填 | 默认值 | 说明 |
|
||||||
|
|------|------|------|--------|------|
|
||||||
|
| `--PROMPT` | `str` | ✅ | - | 发送给 Codex 的任务指令 |
|
||||||
|
| `--cd` | `Path` | ✅ | - | 工作目录根路径 |
|
||||||
|
| `--sandbox` | `Literal` | ❌ | `read-only` | 沙箱策略：`read-only` / `workspace-write` / `danger-full-access` |
|
||||||
|
| `--SESSION_ID` | `UUID` | ❌ | `None` | 会话 ID（None 则开启新会话） |
|
||||||
|
| `--skip-git-repo-check` | `bool` | ❌ | `False` | 是否允许在非 Git 仓库运行 |
|
||||||
|
| `--return-all-messages` | `bool` | ❌ | `False` | 是否返回完整推理信息 |
|
||||||
|
| `--image` | `List[Path]` | ❌ | `None` | 附加图片文件到初始提示词 |
|
||||||
|
| `--model` | `str` | ❌ | `None` | 指定模型（仅在用户明确要求时使用） |
|
||||||
|
| `--yolo` | `bool` | ❌ | `False` | 跳过所有审批与沙箱限制 |
|
||||||
|
|
||||||
|
</details>
|
||||||
|
|
||||||
|
<details>
|
||||||
|
<summary>点击查看使用示例</summary>
|
||||||
|
|
||||||
|
**基础调用：**
|
||||||
|
```bash
|
||||||
|
python scripts/codex_bridge.py --cd "/project" --PROMPT "分析认证流程"
|
||||||
|
```
|
||||||
|
|
||||||
|
**多轮对话：**
|
||||||
|
```bash
|
||||||
|
# 开启新会话
|
||||||
|
python scripts/codex_bridge.py --cd "/project" --PROMPT "审查 login.py 的安全问题"
|
||||||
|
|
||||||
|
# 继续会话
|
||||||
|
python scripts/codex_bridge.py --cd "/project" --SESSION_ID "uuid-from-response" --PROMPT "针对发现的问题提供修复建议"
|
||||||
|
```
|
||||||
|
|
||||||
|
**获取 Diff 原型：**
|
||||||
|
```bash
|
||||||
|
python scripts/codex_bridge.py --cd "/project" --PROMPT "生成添加日志功能的 unified diff"
|
||||||
|
```
|
||||||
|
|
||||||
|
</details>
|
||||||
|
|
||||||
|
### collaborating-with-gemini
|
||||||
|
|
||||||
|
将编码任务委托给 Google Gemini CLI，适用于算法实现、Bug 分析和代码审查。
|
||||||
|
|
||||||
|
<details>
|
||||||
|
<summary>点击查看参数说明</summary>
|
||||||
|
|
||||||
|
| 参数 | 类型 | 必填 | 默认值 | 说明 |
|
||||||
|
|------|------|------|--------|------|
|
||||||
|
| `--PROMPT` | `str` | ✅ | - | 发送给 Gemini 的任务指令 |
|
||||||
|
| `--cd` | `Path` | ✅ | - | 工作目录根路径 |
|
||||||
|
| `--sandbox` | `bool` | ❌ | `False` | 是否启用沙箱模式 |
|
||||||
|
| `--SESSION_ID` | `str` | ❌ | `""` | 会话 ID（空字符串则开启新会话） |
|
||||||
|
| `--return-all-messages` | `bool` | ❌ | `False` | 是否返回完整推理信息 |
|
||||||
|
| `--model` | `str` | ❌ | `None` | 指定模型（仅在用户明确要求时使用） |
|
||||||
|
|
||||||
|
</details>
|
||||||
|
|
||||||
|
<details>
|
||||||
|
<summary>点击查看使用示例</summary>
|
||||||
|
|
||||||
|
**基础调用：**
|
||||||
|
```bash
|
||||||
|
python scripts/gemini_bridge.py --cd "/project" --PROMPT "分析认证流程"
|
||||||
|
```
|
||||||
|
|
||||||
|
**多轮对话：**
|
||||||
|
```bash
|
||||||
|
# 开启新会话
|
||||||
|
python scripts/gemini_bridge.py --cd "/project" --PROMPT "审查 login.py 的安全问题"
|
||||||
|
|
||||||
|
# 继续会话
|
||||||
|
python scripts/gemini_bridge.py --cd "/project" --SESSION_ID "uuid-from-response" --PROMPT "针对发现的问题提供修复建议"
|
||||||
|
```
|
||||||
|
|
||||||
|
</details>
|
||||||
|
|
||||||
|
### 返回值结构
|
||||||
|
|
||||||
|
两个 Skill 均返回统一的 JSON 格式：
|
||||||
|
|
||||||
|
**成功时：**
|
||||||
|
```json
|
||||||
|
{
|
||||||
|
"success": true,
|
||||||
|
"SESSION_ID": "550e8400-e29b-41d4-a716-446655440000",
|
||||||
|
"agent_messages": "模型回复内容...",
|
||||||
|
"all_messages": []
|
||||||
|
}
|
||||||
|
```
|
||||||
|
|
||||||
|
**失败时：**
|
||||||
|
```json
|
||||||
|
{
|
||||||
|
"success": false,
|
||||||
|
"error": "错误信息描述"
|
||||||
|
}
|
||||||
|
```
|
||||||
|
|
||||||
|
---
|
||||||
|
|
||||||
|
## 四、FAQ
|
||||||
|
|
||||||
|
<details>
|
||||||
|
<summary>Q1: 什么是 Agent Skills？</summary>
|
||||||
|
|
||||||
|
Agent Skills 是 Anthropic 推出的模块化能力扩展机制。每个 Skill 包含指令、元数据和可选资源（脚本、模板），Claude 会在相关任务时自动使用。Skills 采用渐进式加载，仅在需要时加载相关内容，有效管理上下文窗口。
|
||||||
|
|
||||||
|
</details>
|
||||||
|
|
||||||
|
<details>
|
||||||
|
<summary>Q2: 是否需要额外付费？</summary>
|
||||||
|
|
||||||
|
**Skills 本身完全免费开源。** 但使用协作模型（Codex、Gemini）可能产生相应的 API 费用，具体取决于各平台的定价策略。
|
||||||
|
|
||||||
|
</details>
|
||||||
|
|
||||||
|
<details>
|
||||||
|
<summary>Q3: 多轮对话时 SESSION_ID 有什么作用？</summary>
|
||||||
|
|
||||||
|
`SESSION_ID` 用于维护会话上下文。首次调用时返回一个新的 SESSION_ID，后续调用传入该 ID 可继续之前的对话，模型会记住之前的交互内容。每个 SESSION_ID 对应独立的会话，不同会话之间完全隔离。
|
||||||
|
|
||||||
|
</details>
|
||||||
|
|
||||||
|
<details>
|
||||||
|
<summary>Q4: 如何确保代码安全？</summary>
|
||||||
|
|
||||||
|
建议始终使用 `read-only` 沙箱策略，并在 PROMPT 中明确要求模型返回 `unified diff patch` 而非直接修改文件。外部模型生成的代码仅作为参考原型，最终实现应由 Claude 重构确认。
|
||||||
|
|
||||||
|
</details>
|
||||||
|
|
||||||
|
---
|
||||||
|
|
||||||
|
## 🤝 贡献指南
|
||||||
|
|
||||||
|
<details>
|
||||||
|
<summary>我们欢迎所有形式的贡献！</summary>
|
||||||
|
|
||||||
|
### 开发环境配置
|
||||||
|
|
||||||
|
```bash
|
||||||
|
# 克隆仓库（含子模块）
|
||||||
|
git clone --recurse-submodules https://github.com/GuDaStudio/skills.git
|
||||||
|
cd skills
|
||||||
|
```
|
||||||
|
|
||||||
|
### 提交规范
|
||||||
|
|
||||||
|
- 遵循 [Conventional Commits](https://www.conventionalcommits.org/)
|
||||||
|
- 提交测试用例
|
||||||
|
- 更新相关文档
|
||||||
|
|
||||||
|
</details>
|
||||||
|
|
||||||
|
---
|
||||||
|
|
||||||
|
## 📄 许可证
|
||||||
|
|
||||||
|
本项目采用 [MIT License](LICENSE) 开源协议。
|
||||||
|
|
||||||
|
Copyright (c) 2025 [guda.studio](mailto:gudaclaude@gmail.com)
|
||||||
|
|
||||||
|
---
|
||||||
|
|
||||||
|
<div align="center">
|
||||||
|
|
||||||
|
## 用 🌟 为本项目助力~
|
||||||
|
[](https://www.star-history.com/#GuDaStudio/skills&type=date&legend=top-left)
|
||||||
|
|
||||||
|
</div>
|
||||||
369
docs/README_EN.md
Normal file
369
docs/README_EN.md
Normal file
@@ -0,0 +1,369 @@
|
|||||||
|
|
||||||
|
|
||||||
|
<div align="center">
|
||||||
|
|
||||||
|
**Agent Skills Collection — Seamless Collaboration Between Claude and Multi-Model/Tools**
|
||||||
|
|
||||||
|
[](https://opensource.org/licenses/MIT) [](https://claude.ai/code) [](https://x.com/intent/tweet?text=GudaCC%20Skills：Claude%20Code%20Agent%20Skills%20Collection%20https://github.com/GuDaStudio/skills%20%23AI%20%23Coding%20%23ClaudeCode) [](https://www.facebook.com/sharer/sharer.php?u=https://github.com/GuDaStudio/skills) [](https://t.me/share/url?url=https://github.com/GuDaStudio/skills&text=GudaCC%20Skills：Claude%20Code%20Agent%20Skills%20Collection)
|
||||||
|
|
||||||
|
Star us on GitHub — your support means a lot! 🙏😊
|
||||||
|
|
||||||
|
[简体中文](../README.md) | English
|
||||||
|
|
||||||
|
</div>
|
||||||
|
|
||||||
|
---
|
||||||
|
|
||||||
|
## Introduction
|
||||||
|
|
||||||
|
**Agent Skills** is a modular capability extension mechanism [introduced by Anthropic](https://platform.claude.com/docs/en/agents-and-tools/agent-skills/overview) that enables LLMs to load domain-specific knowledge and workflows on demand. This repository contains a collection of Agent Skills developed by **GudaStudio**, enabling seamless collaboration between Claude and other AI models/tools.
|
||||||
|
|
||||||
|
### Available Skills
|
||||||
|
|
||||||
|
| Skill | Description | Collaborating Model |
|
||||||
|
|-------|-------------|---------------------|
|
||||||
|
| [collaborating-with-codex](../collaborating-with-codex) | Delegates coding tasks to Codex CLI for prototyping, debugging, and code review | OpenAI Codex |
|
||||||
|
| [collaborating-with-gemini](../collaborating-with-gemini) | Delegates coding tasks to Gemini CLI for prototyping, debugging, and code review | Google Gemini |
|
||||||
|
|
||||||
|
---
|
||||||
|
|
||||||
|
## Quick Start
|
||||||
|
|
||||||
|
### 0. Prerequisites
|
||||||
|
|
||||||
|
Ensure the following tools are installed and configured:
|
||||||
|
|
||||||
|
- [Claude Code](https://docs.claude.com/docs/claude-code) (v2.0.56+)
|
||||||
|
- [Codex CLI](https://github.com/openai/codex) (if using collaborating-with-codex)
|
||||||
|
- [Gemini CLI](https://github.com/google-gemini/gemini-cli) (if using collaborating-with-gemini)
|
||||||
|
|
||||||
|
### 1. Clone Repository
|
||||||
|
|
||||||
|
```bash
|
||||||
|
# Clone to any location
|
||||||
|
git clone --recurse-submodules https://github.com/GuDaStudio/skills
|
||||||
|
cd skills
|
||||||
|
```
|
||||||
|
|
||||||
|
### 2. Install Skills
|
||||||
|
|
||||||
|
The install script provides flexible options for scope and target location.
|
||||||
|
|
||||||
|
**List available Skills:**
|
||||||
|
|
||||||
|
```bash
|
||||||
|
./install.sh --list
|
||||||
|
```
|
||||||
|
|
||||||
|
**Option 1: Install All Skills**
|
||||||
|
|
||||||
|
```bash
|
||||||
|
# User-level installation (available for all projects)
|
||||||
|
./install.sh --user --all
|
||||||
|
|
||||||
|
# Project-level installation (current project only, run from project root)
|
||||||
|
./install.sh --project --all
|
||||||
|
```
|
||||||
|
|
||||||
|
**Option 2: Selective Installation**
|
||||||
|
|
||||||
|
```bash
|
||||||
|
# Install only collaborating-with-codex
|
||||||
|
./install.sh --user --skill collaborating-with-codex
|
||||||
|
|
||||||
|
# Install multiple specific Skills
|
||||||
|
./install.sh --user -s collaborating-with-codex -s collaborating-with-gemini
|
||||||
|
```
|
||||||
|
|
||||||
|
**Option 3: Custom Installation Path**
|
||||||
|
|
||||||
|
```bash
|
||||||
|
./install.sh --target /your/custom/path --all
|
||||||
|
```
|
||||||
|
|
||||||
|
<details>
|
||||||
|
<summary>Click to view full parameter reference</summary>
|
||||||
|
|
||||||
|
| Parameter | Short | Description |
|
||||||
|
|-----------|-------|-------------|
|
||||||
|
| `--user` | `-u` | Install to user-level directory (`~/.claude/skills/`) |
|
||||||
|
| `--project` | `-p` | Install to project-level directory (`./.claude/skills/`) |
|
||||||
|
| `--target <path>` | `-t` | Install to custom path |
|
||||||
|
| `--all` | `-a` | Install all available Skills |
|
||||||
|
| `--skill <name>` | `-s` | Install specific Skill (can be used multiple times) |
|
||||||
|
| `--list` | `-l` | List all available Skills |
|
||||||
|
| `--help` | `-h` | Show help message |
|
||||||
|
|
||||||
|
</details>
|
||||||
|
|
||||||
|
### 3. Verify Installation
|
||||||
|
|
||||||
|
Skills will auto-load when Claude Code starts. Verify by:
|
||||||
|
|
||||||
|
- Ask Claude to list all skills: `list all available SKILLs please`
|
||||||
|
- Claude should recognize and invoke `collaborating-with-codex` and `collaborating-with-gemini`
|
||||||
|
|
||||||
|
|
||||||
|
### 4. (🎊 Optional, Highly Recommended) Configure Prompts
|
||||||
|
|
||||||
|
To maximize the effectiveness of this SKILLS collection with Claude Code CLI, we strongly recommend adding the following prompts to `~/.claude/CLAUDE.md`.
|
||||||
|
|
||||||
|
````markdown
|
||||||
|
# CLAUDE.md
|
||||||
|
|
||||||
|
## 0. Global Protocols
|
||||||
|
|
||||||
|
All operations must strictly follow these system constraints:
|
||||||
|
|
||||||
|
- **Interaction Language**: Use **English** when interacting with tools or models; use **Chinese** for user output.
|
||||||
|
- **Multi-turn Conversations**: If tools or models return a persistent conversation field like `SESSION_ID`, indicating multi-turn support, record this field and **mandatory think** in subsequent calls whether to continue the conversation. For example, Codex/Gemini may interrupt sessions due to internal tool calls; if you don't get the needed response, continue the conversation.
|
||||||
|
- **Sandbox Security**: Codex/Gemini are **strictly forbidden** from writing to the filesystem. All code retrieval must **explicitly require** in the PROMPT to return `Unified Diff Patch`; Codex/Gemini are forbidden from making any real modifications.
|
||||||
|
- **Code Sovereignty**: Code generated by external models serves only as logical reference (Prototype); final delivered code **must be refactored** to ensure no redundancy and enterprise-grade standards.
|
||||||
|
- **Style Definition**: Overall code style is **always positioned** as lean, efficient, and zero redundancy. This requirement also applies to comments and documentation, which strictly follow the core principle of **not forming unless necessary**.
|
||||||
|
- **Only make targeted changes for requirements**, strictly forbidden to affect user's other existing functionality.
|
||||||
|
- Codex/Gemini interaction methods are provided as SKILLs, **mandatory to actively view and invoke**. Both require longer invocation time (system default "BASH_DEFAULT_TIMEOUT_MS": "300000"), just wait patiently.
|
||||||
|
- When parallelizable tasks are detected, execute in parallel as much as possible. For example, when multiple SKILL-related bash commands exist, use `run in background` to **immediately** suspend long-running programs to execute the next command, achieving parallelism.
|
||||||
|
- Mandatory to follow all Phases in 1. Workflow, strictly forbidden to omit any process.
|
||||||
|
|
||||||
|
## 1. Workflow
|
||||||
|
|
||||||
|
### Phase 1: Full Context Retrieval (Auggie Interface)
|
||||||
|
|
||||||
|
**Execution Condition**: Before generating any suggestions or code.
|
||||||
|
|
||||||
|
1. **Tool Invocation**: Call `mcp__auggie-mcp__codebase-retrieval`.
|
||||||
|
2. **Retrieval Strategy**:
|
||||||
|
- Forbidden to answer based on assumptions.
|
||||||
|
- Use Natural Language (NL) to construct semantic queries (Where/What/How).
|
||||||
|
- **Completeness Check**: Must obtain complete definitions and signatures of relevant classes, functions, and variables. If context is insufficient, trigger recursive retrieval.
|
||||||
|
3. **Requirements Alignment**: If requirements still have ambiguity after retrieval, **must** output a list of guiding questions to the user until requirement boundaries are clear (no omissions, no redundancy).
|
||||||
|
|
||||||
|
### Phase 2: Multi-Model Collaborative Analysis (Analysis & Strategy)
|
||||||
|
|
||||||
|
**Execution Condition**: After context is ready, before coding begins.
|
||||||
|
|
||||||
|
1. **Distribute Input**: Distribute user's **original requirements** (without preset opinions) to Codex and Gemini. Note that Codex/Gemini both have complete CLI systems, so **no need to provide excessive context**.
|
||||||
|
2. **Solution Iteration**:
|
||||||
|
- Require models to provide multi-perspective solutions.
|
||||||
|
- Trigger **Cross-validation**: Integrate ideas from all parties, iterate and optimize, perform logical deduction and complement advantages/disadvantages during the process, until generating a Step-by-step implementation plan with no logical gaps.
|
||||||
|
3. **User Confirmation**: Present the final implementation plan (with moderate pseudo-code) to the user.
|
||||||
|
|
||||||
|
### Phase 3: Prototype Acquisition (Prototyping)
|
||||||
|
|
||||||
|
**Execution Condition**: After implementation plan is confirmed. Route based on task type:
|
||||||
|
|
||||||
|
- **Route A: Frontend/UI/Styles (Gemini Kernel)**
|
||||||
|
- **Limitation**: Context < 32k. Gemini has deficiencies in understanding backend logic; its responses need objective review.
|
||||||
|
- **Instruction**: Request CSS/React/Vue prototypes. Use this as the final frontend design prototype and visual baseline.
|
||||||
|
- **Route B: Backend/Logic/Algorithm (Codex Kernel)**
|
||||||
|
- **Capability**: Leverage its logical computation and debugging abilities.
|
||||||
|
- **Instruction**: Request logic implementation prototypes.
|
||||||
|
- **Universal Constraint**: In any communication with Codex/Gemini, **must explicitly require** in the Prompt to return `Unified Diff Patch`; Codex/Gemini are forbidden from making any real modifications.
|
||||||
|
|
||||||
|
### Phase 4: Coding Implementation (Implementation)
|
||||||
|
|
||||||
|
**Execution Guidelines**:
|
||||||
|
|
||||||
|
1. **Logic Refactoring**: Based on Phase 3 prototypes, remove redundancy, **rewrite** into highly readable, highly maintainable, enterprise-release-grade code.
|
||||||
|
2. **Documentation Standards**: Do not generate comments and documentation unless necessary; code should be self-explanatory.
|
||||||
|
3. **Minimal Scope**: Changes limited to requirement scope, **mandatory review** whether changes introduce side effects and make targeted corrections.
|
||||||
|
|
||||||
|
### Phase 5: Audit & Delivery (Audit & Delivery)
|
||||||
|
|
||||||
|
1. **Automatic Audit**: After changes take effect, **mandatory immediate invocation** of both Codex and Gemini for Code Review simultaneously, and perform integrated fixes.
|
||||||
|
- Check items: Logic correctness, requirement coverage, potential bugs.
|
||||||
|
2. **Delivery**: Feedback to user after audit passes.
|
||||||
|
|
||||||
|
## 2. Resource Matrix
|
||||||
|
|
||||||
|
This matrix defines the **mandatory** resource invocation strategy for each phase. Claude as the **Orchestrator** must strictly schedule external resources according to the current Workflow phase per the following specifications.
|
||||||
|
|
||||||
|
| Workflow Phase | Functionality | Designated Model / Tool | Input Strategy (Prompting) | Strict Output Constraints | Critical Constraints & Behavior |
|
||||||
|
| :----------------------- | :---------------------- | :--------------------------------------- | :----------------------------------------------------------- | :-------------------------------------------------- | :----------------------------------------------------------- |
|
||||||
|
| **Phase 1** | **Context Retrieval** | **Auggie** (`mcp__auggie`) | **Natural Language (English)**<br>Focus on: *What, Where, How* | **Raw Code / Definitions**<br>(Complete Signatures) | • **Forbidden:** `grep` / keyword search.<br>• **Mandatory:** Recursive retrieval until context is complete. |
|
||||||
|
| **Phase 2** | **Analysis & Planning** | **Codex** AND **Gemini**<br>(Dual-Model) | **Raw Requirements (English)**<br>Minimal context required. | **Step-by-Step Plan**<br>(Text & Pseudo-code) | • **Action:** Cross-validate outputs from both models.<br>• **Goal:** Eliminate logic gaps before coding starts. |
|
||||||
|
| **Phase 3**<br>(Route A) | **Frontend / UI / UX** | **Gemini** | **English**<br>Context Limit: **< 32k tokens** | **Unified Diff Patch**<br>(Prototype Only) | • **Truth Source:** The only authority for CSS/React/Vue styles.<br>• **Warning:** Ignore its backend logic suggestions. |
|
||||||
|
| **Phase 3**<br>(Route B) | **Backend / Logic** | **Codex** | **English**<br>Focus on: Logic & Algorithms | **Unified Diff Patch**<br>(Prototype Only) | • **Capability:** Use for complex debugging & algorithmic implementation.<br>• **Security:** **NO** file system write access allowed. |
|
||||||
|
| **Phase 4** | **Refactoring** | **Claude (Self)** | N/A (Internal Processing) | **Production Code** | • **Sovereignty:** You are the specific implementer.<br>• **Style:** Clean, efficient, no redundancy. Minimal comments. |
|
||||||
|
| **Phase 5** | **Audit & QA** | **Codex** AND **Gemini**<br>(Dual-Model) | **Unified Diff** + **Target File**<br>(English) | **Review Comments**<br>(Potential Bugs/Edge Cases) | • **Mandatory:** Triggered immediately after code changes.<br>• **Action:** Synthesize feedback into a final fix. |
|
||||||
|
````
|
||||||
|
|
||||||
|
---
|
||||||
|
|
||||||
|
## Skills Reference
|
||||||
|
|
||||||
|
### collaborating-with-codex
|
||||||
|
|
||||||
|
Delegates coding tasks to OpenAI Codex CLI for algorithm implementation, bug analysis, and code review.
|
||||||
|
|
||||||
|
<details>
|
||||||
|
<summary>Click to view parameters</summary>
|
||||||
|
|
||||||
|
| Parameter | Type | Required | Default | Description |
|
||||||
|
|-----------|------|----------|---------|-------------|
|
||||||
|
| `--PROMPT` | `str` | ✅ | - | Task instruction for Codex |
|
||||||
|
| `--cd` | `Path` | ✅ | - | Working directory root path |
|
||||||
|
| `--sandbox` | `Literal` | ❌ | `read-only` | Sandbox policy: `read-only` / `workspace-write` / `danger-full-access` |
|
||||||
|
| `--SESSION_ID` | `UUID` | ❌ | `None` | Session ID (None starts new session) |
|
||||||
|
| `--skip-git-repo-check` | `bool` | ❌ | `False` | Allow running in non-Git repository |
|
||||||
|
| `--return-all-messages` | `bool` | ❌ | `False` | Return complete reasoning information |
|
||||||
|
| `--image` | `List[Path]` | ❌ | `None` | Attach image files to initial prompt |
|
||||||
|
| `--model` | `str` | ❌ | `None` | Specify model (use only when explicitly requested) |
|
||||||
|
| `--yolo` | `bool` | ❌ | `False` | Skip all approvals and sandbox restrictions |
|
||||||
|
|
||||||
|
</details>
|
||||||
|
|
||||||
|
<details>
|
||||||
|
<summary>Click to view usage examples</summary>
|
||||||
|
|
||||||
|
**Basic call:**
|
||||||
|
```bash
|
||||||
|
python scripts/codex_bridge.py --cd "/project" --PROMPT "Analyze authentication flow"
|
||||||
|
```
|
||||||
|
|
||||||
|
**Multi-turn conversation:**
|
||||||
|
```bash
|
||||||
|
# Start new session
|
||||||
|
python scripts/codex_bridge.py --cd "/project" --PROMPT "Review login.py for security issues"
|
||||||
|
|
||||||
|
# Continue session
|
||||||
|
python scripts/codex_bridge.py --cd "/project" --SESSION_ID "uuid-from-response" --PROMPT "Provide fix suggestions for identified issues"
|
||||||
|
```
|
||||||
|
|
||||||
|
**Get diff prototype:**
|
||||||
|
```bash
|
||||||
|
python scripts/codex_bridge.py --cd "/project" --PROMPT "Generate unified diff for adding logging feature"
|
||||||
|
```
|
||||||
|
|
||||||
|
</details>
|
||||||
|
|
||||||
|
### collaborating-with-gemini
|
||||||
|
|
||||||
|
Delegates coding tasks to Google Gemini CLI for algorithm implementation, bug analysis, and code review.
|
||||||
|
|
||||||
|
<details>
|
||||||
|
<summary>Click to view parameters</summary>
|
||||||
|
|
||||||
|
| Parameter | Type | Required | Default | Description |
|
||||||
|
|-----------|------|----------|---------|-------------|
|
||||||
|
| `--PROMPT` | `str` | ✅ | - | Task instruction for Gemini |
|
||||||
|
| `--cd` | `Path` | ✅ | - | Working directory root path |
|
||||||
|
| `--sandbox` | `bool` | ❌ | `False` | Enable sandbox mode |
|
||||||
|
| `--SESSION_ID` | `str` | ❌ | `""` | Session ID (empty string starts new session) |
|
||||||
|
| `--return-all-messages` | `bool` | ❌ | `False` | Return complete reasoning information |
|
||||||
|
| `--model` | `str` | ❌ | `None` | Specify model (use only when explicitly requested) |
|
||||||
|
|
||||||
|
</details>
|
||||||
|
|
||||||
|
<details>
|
||||||
|
<summary>Click to view usage examples</summary>
|
||||||
|
|
||||||
|
**Basic call:**
|
||||||
|
```bash
|
||||||
|
python scripts/gemini_bridge.py --cd "/project" --PROMPT "Analyze authentication flow"
|
||||||
|
```
|
||||||
|
|
||||||
|
**Multi-turn conversation:**
|
||||||
|
```bash
|
||||||
|
# Start new session
|
||||||
|
python scripts/gemini_bridge.py --cd "/project" --PROMPT "Review login.py for security issues"
|
||||||
|
|
||||||
|
# Continue session
|
||||||
|
python scripts/gemini_bridge.py --cd "/project" --SESSION_ID "uuid-from-response" --PROMPT "Provide fix suggestions for identified issues"
|
||||||
|
```
|
||||||
|
|
||||||
|
</details>
|
||||||
|
|
||||||
|
### Return Value Structure
|
||||||
|
|
||||||
|
Both Skills return a unified JSON format:
|
||||||
|
|
||||||
|
**On success:**
|
||||||
|
```json
|
||||||
|
{
|
||||||
|
"success": true,
|
||||||
|
"SESSION_ID": "550e8400-e29b-41d4-a716-446655440000",
|
||||||
|
"agent_messages": "Model response content...",
|
||||||
|
"all_messages": []
|
||||||
|
}
|
||||||
|
```
|
||||||
|
|
||||||
|
**On failure:**
|
||||||
|
```json
|
||||||
|
{
|
||||||
|
"success": false,
|
||||||
|
"error": "Error description"
|
||||||
|
}
|
||||||
|
```
|
||||||
|
|
||||||
|
---
|
||||||
|
|
||||||
|
## FAQ
|
||||||
|
|
||||||
|
<details>
|
||||||
|
<summary>Q1: What are Agent Skills?</summary>
|
||||||
|
|
||||||
|
Agent Skills is a modular capability extension mechanism introduced by Anthropic. Each Skill contains instructions, metadata, and optional resources (scripts, templates). Claude automatically uses them for relevant tasks. Skills use progressive loading, only loading relevant content when needed, effectively managing the context window.
|
||||||
|
|
||||||
|
</details>
|
||||||
|
|
||||||
|
<details>
|
||||||
|
<summary>Q2: Are there any additional costs?</summary>
|
||||||
|
|
||||||
|
**Skills themselves are completely free and open source.** However, using collaborating models (Codex, Gemini) may incur API fees depending on each platform's pricing policy.
|
||||||
|
|
||||||
|
</details>
|
||||||
|
|
||||||
|
<details>
|
||||||
|
<summary>Q3: What is the SESSION_ID for in multi-turn conversations?</summary>
|
||||||
|
|
||||||
|
`SESSION_ID` maintains conversation context. The first call returns a new SESSION_ID, and subsequent calls with this ID continue the previous conversation. The model remembers prior interactions. Each SESSION_ID corresponds to an independent session, completely isolated from others.
|
||||||
|
|
||||||
|
</details>
|
||||||
|
|
||||||
|
<details>
|
||||||
|
<summary>Q4: How to ensure code security?</summary>
|
||||||
|
|
||||||
|
Always use the `read-only` sandbox policy and explicitly request the model to return `unified diff patch` in the PROMPT rather than directly modifying files. Code generated by external models should only serve as reference prototypes; final implementation should be refactored and confirmed by Claude.
|
||||||
|
|
||||||
|
</details>
|
||||||
|
|
||||||
|
---
|
||||||
|
|
||||||
|
## Contributing
|
||||||
|
|
||||||
|
<details>
|
||||||
|
<summary>We welcome all forms of contributions!</summary>
|
||||||
|
|
||||||
|
### Development Setup
|
||||||
|
|
||||||
|
```bash
|
||||||
|
# Clone repository (with submodules)
|
||||||
|
git clone --recurse-submodules https://github.com/GuDaStudio/skills.git
|
||||||
|
cd skills
|
||||||
|
```
|
||||||
|
|
||||||
|
### Commit Guidelines
|
||||||
|
|
||||||
|
- Follow [Conventional Commits](https://www.conventionalcommits.org/)
|
||||||
|
- Include test cases
|
||||||
|
- Update relevant documentation
|
||||||
|
|
||||||
|
</details>
|
||||||
|
|
||||||
|
---
|
||||||
|
|
||||||
|
## License
|
||||||
|
|
||||||
|
This project is open-sourced under the [MIT License](../LICENSE).
|
||||||
|
|
||||||
|
Copyright (c) 2025 [guda.studio](mailto:gudaclaude@gmail.com)
|
||||||
|
|
||||||
|
---
|
||||||
|
|
||||||
|
<div align="center">
|
||||||
|
|
||||||
|
## Star this project!
|
||||||
|
[](https://www.star-history.com/#GuDaStudio/skills&type=date&legend=top-left)
|
||||||
|
|
||||||
|
</div>
|
||||||
BIN
imgs/title.png
Normal file
BIN
imgs/title.png
Normal file
Binary file not shown.
|
After Width: | Height: | Size: 1.7 MiB |
BIN
imgs/verify.png
Normal file
BIN
imgs/verify.png
Normal file
Binary file not shown.
|
After Width: | Height: | Size: 79 KiB |
191
install.sh
Normal file
191
install.sh
Normal file
@@ -0,0 +1,191 @@
|
|||||||
|
#!/bin/bash
|
||||||
|
# GudaCC Skills Installer
|
||||||
|
# https://github.com/GuDaStudio/skills
|
||||||
|
|
||||||
|
set -e
|
||||||
|
|
||||||
|
# Colors
|
||||||
|
RED='\033[0;31m'
|
||||||
|
GREEN='\033[0;32m'
|
||||||
|
YELLOW='\033[1;33m'
|
||||||
|
BLUE='\033[0;34m'
|
||||||
|
NC='\033[0m'
|
||||||
|
|
||||||
|
# Available skills
|
||||||
|
AVAILABLE_SKILLS=("collaborating-with-codex" "collaborating-with-gemini")
|
||||||
|
|
||||||
|
# Script directory
|
||||||
|
SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
|
||||||
|
|
||||||
|
usage() {
|
||||||
|
echo -e "${BLUE}GudaCC Skills Installer${NC}"
|
||||||
|
echo ""
|
||||||
|
echo "Usage: $0 [OPTIONS]"
|
||||||
|
echo ""
|
||||||
|
echo "Options:"
|
||||||
|
echo " -u, --user Install to user-level (~/.claude/skills/)"
|
||||||
|
echo " -p, --project Install to project-level (./.claude/skills/)"
|
||||||
|
echo " -t, --target <path> Install to custom target path"
|
||||||
|
echo " -a, --all Install all available skills"
|
||||||
|
echo " -s, --skill <name> Install specific skill (can be used multiple times)"
|
||||||
|
echo " -l, --list List available skills"
|
||||||
|
echo " -h, --help Show this help message"
|
||||||
|
echo ""
|
||||||
|
echo "Examples:"
|
||||||
|
echo " $0 --user --all # Install all skills to user-level"
|
||||||
|
echo " $0 --project --all # Install all skills to current project"
|
||||||
|
echo " $0 --user -s collaborating-with-codex"
|
||||||
|
echo " $0 --target /custom/path --all"
|
||||||
|
echo ""
|
||||||
|
echo "Available skills:"
|
||||||
|
for skill in "${AVAILABLE_SKILLS[@]}"; do
|
||||||
|
echo " - $skill"
|
||||||
|
done
|
||||||
|
}
|
||||||
|
|
||||||
|
list_skills() {
|
||||||
|
echo -e "${BLUE}Available Skills:${NC}"
|
||||||
|
echo ""
|
||||||
|
for skill in "${AVAILABLE_SKILLS[@]}"; do
|
||||||
|
if [ -d "$SCRIPT_DIR/$skill" ]; then
|
||||||
|
echo -e " ${GREEN}✓${NC} $skill"
|
||||||
|
else
|
||||||
|
echo -e " ${RED}✗${NC} $skill (not found in source)"
|
||||||
|
fi
|
||||||
|
done
|
||||||
|
}
|
||||||
|
|
||||||
|
install_skill() {
|
||||||
|
local skill=$1
|
||||||
|
local target_dir=$2
|
||||||
|
local source_dir="$SCRIPT_DIR/$skill"
|
||||||
|
local dest_dir="$target_dir/$skill"
|
||||||
|
|
||||||
|
if [ ! -d "$source_dir" ]; then
|
||||||
|
echo -e "${RED}Error: Skill '$skill' not found in source directory${NC}"
|
||||||
|
return 1
|
||||||
|
fi
|
||||||
|
|
||||||
|
echo -e "${BLUE}Installing${NC} $skill -> $dest_dir"
|
||||||
|
|
||||||
|
# Create target directory
|
||||||
|
mkdir -p "$target_dir"
|
||||||
|
|
||||||
|
# Remove existing installation
|
||||||
|
if [ -d "$dest_dir" ]; then
|
||||||
|
echo -e "${YELLOW} Removing existing installation...${NC}"
|
||||||
|
rm -rf "$dest_dir"
|
||||||
|
fi
|
||||||
|
|
||||||
|
# Copy skill directory
|
||||||
|
cp -r "$source_dir" "$dest_dir"
|
||||||
|
|
||||||
|
# Remove .git file if it's a submodule reference
|
||||||
|
if [ -f "$dest_dir/.git" ]; then
|
||||||
|
rm "$dest_dir/.git"
|
||||||
|
fi
|
||||||
|
|
||||||
|
echo -e "${GREEN} ✓ Installed${NC}"
|
||||||
|
}
|
||||||
|
|
||||||
|
# Parse arguments
|
||||||
|
TARGET_PATH=""
|
||||||
|
INSTALL_ALL=false
|
||||||
|
SELECTED_SKILLS=()
|
||||||
|
SCOPE=""
|
||||||
|
|
||||||
|
while [[ $# -gt 0 ]]; do
|
||||||
|
case $1 in
|
||||||
|
-u|--user)
|
||||||
|
SCOPE="user"
|
||||||
|
TARGET_PATH="$HOME/.claude/skills"
|
||||||
|
shift
|
||||||
|
;;
|
||||||
|
-p|--project)
|
||||||
|
SCOPE="project"
|
||||||
|
TARGET_PATH="./.claude/skills"
|
||||||
|
shift
|
||||||
|
;;
|
||||||
|
-t|--target)
|
||||||
|
TARGET_PATH="$2"
|
||||||
|
shift 2
|
||||||
|
;;
|
||||||
|
-a|--all)
|
||||||
|
INSTALL_ALL=true
|
||||||
|
shift
|
||||||
|
;;
|
||||||
|
-s|--skill)
|
||||||
|
SELECTED_SKILLS+=("$2")
|
||||||
|
shift 2
|
||||||
|
;;
|
||||||
|
-l|--list)
|
||||||
|
list_skills
|
||||||
|
exit 0
|
||||||
|
;;
|
||||||
|
-h|--help)
|
||||||
|
usage
|
||||||
|
exit 0
|
||||||
|
;;
|
||||||
|
*)
|
||||||
|
echo -e "${RED}Unknown option: $1${NC}"
|
||||||
|
usage
|
||||||
|
exit 1
|
||||||
|
;;
|
||||||
|
esac
|
||||||
|
done
|
||||||
|
|
||||||
|
# Validate arguments
|
||||||
|
if [ -z "$TARGET_PATH" ]; then
|
||||||
|
echo -e "${RED}Error: Please specify installation target (-u, -p, or -t)${NC}"
|
||||||
|
echo ""
|
||||||
|
usage
|
||||||
|
exit 1
|
||||||
|
fi
|
||||||
|
|
||||||
|
if [ "$INSTALL_ALL" = false ] && [ ${#SELECTED_SKILLS[@]} -eq 0 ]; then
|
||||||
|
echo -e "${RED}Error: Please specify skills to install (-a or -s)${NC}"
|
||||||
|
echo ""
|
||||||
|
usage
|
||||||
|
exit 1
|
||||||
|
fi
|
||||||
|
|
||||||
|
# Determine skills to install
|
||||||
|
if [ "$INSTALL_ALL" = true ]; then
|
||||||
|
SKILLS_TO_INSTALL=("${AVAILABLE_SKILLS[@]}")
|
||||||
|
else
|
||||||
|
SKILLS_TO_INSTALL=("${SELECTED_SKILLS[@]}")
|
||||||
|
fi
|
||||||
|
|
||||||
|
# Validate selected skills
|
||||||
|
for skill in "${SKILLS_TO_INSTALL[@]}"; do
|
||||||
|
valid=false
|
||||||
|
for available in "${AVAILABLE_SKILLS[@]}"; do
|
||||||
|
if [ "$skill" = "$available" ]; then
|
||||||
|
valid=true
|
||||||
|
break
|
||||||
|
fi
|
||||||
|
done
|
||||||
|
if [ "$valid" = false ]; then
|
||||||
|
echo -e "${RED}Error: Unknown skill '$skill'${NC}"
|
||||||
|
echo "Available skills: ${AVAILABLE_SKILLS[*]}"
|
||||||
|
exit 1
|
||||||
|
fi
|
||||||
|
done
|
||||||
|
|
||||||
|
# Install
|
||||||
|
echo -e "${BLUE}━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━${NC}"
|
||||||
|
echo -e "${BLUE}GudaCC Skills Installer${NC}"
|
||||||
|
echo -e "${BLUE}━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━${NC}"
|
||||||
|
echo ""
|
||||||
|
echo -e "Target: ${GREEN}$TARGET_PATH${NC}"
|
||||||
|
echo -e "Skills: ${GREEN}${SKILLS_TO_INSTALL[*]}${NC}"
|
||||||
|
echo ""
|
||||||
|
|
||||||
|
for skill in "${SKILLS_TO_INSTALL[@]}"; do
|
||||||
|
install_skill "$skill" "$TARGET_PATH"
|
||||||
|
done
|
||||||
|
|
||||||
|
echo ""
|
||||||
|
echo -e "${GREEN}━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━${NC}"
|
||||||
|
echo -e "${GREEN}Installation complete!${NC}"
|
||||||
|
echo -e "${GREEN}━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━${NC}"
|
||||||
Reference in New Issue
Block a user