<rss xmlns:atom="http://www.w3.org/2005/Atom" version="2.0">
  <channel>
    <title>opencode - 标签 - cfanzp学习笔记</title>
    <link>https://cfanzp008.github.io/tags/opencode/</link>
    <description>opencode - 标签 - cfanzp学习笔记</description>
    <generator>Hugo -- gohugo.io</generator><language>zh-CN</language><managingEditor>cfan.zp@qq.com (cfanzp)</managingEditor>
      <webMaster>cfan.zp@qq.com (cfanzp)</webMaster><lastBuildDate>Wed, 08 Apr 2026 18:04:31 &#43;0800</lastBuildDate><atom:link href="https://cfanzp008.github.io/tags/opencode/" rel="self" type="application/rss+xml" /><item>
  <title>Superpowers 自动化使用指南</title>
  <link>https://cfanzp008.github.io/superpowers-automation-guide/</link>
  <pubDate>Wed, 08 Apr 2026 18:04:31 &#43;0800</pubDate>
  <author>作者</author>
  <guid>https://cfanzp008.github.io/superpowers-automation-guide/</guid>
  <description><![CDATA[Superpowers 自动化使用指南 什么是 Superpowers Superpowers 是 OpenCode 平台推出的一套 AI 技能框架，旨在为 AI Agent 提供专家级能力。通过 Superpowers，用户可以将特定领域的知识和工作流封装为可重用的技能（Skills），让 AI 在执行任务时能够自动调用这些技能。
Superpowers 的核心理念是「让 AI 具备专业领域知识」，就像为 AI 配备了一个随身携带的专家团队。
Superpowers 架构 技能类型 Superpowers 将技能分为两类：
Rigid Skills（刚性技能）：严格遵循预设流程，如 TDD（测试驱动开发）、Debugging
使用时必须完全按照规范执行 不允许随意调整流程 Flexible Skills（柔性技能）：可以灵活适应上下文，如 Patterns
可以根据实际情况调整实现方式 需要理解技能背后的原理 技能层级 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 Superpowers ├── 基础技能（Built-in） │ ├── playwright │ ├── frontend-ui-ux │ ├── git-master │ ├── dev-browser │ ├── review-work │ └── ai-slop-remover │ ├── 用户技能（User-installed） │ ├── doc-coauthoring │ ├── blog-do-publish │ ├── pdf │ ├── docx │ ├── xlsx │ ├── pptx │ ├── mcp-builder │ └── article-writer │ └── 超级技能（Superpowers） ├── using-superpowers ├── brainstorming ├── test-driven-development ├── systematic-debugging ├── verification-before-completion ├── requesting-code-review ├── receiving-code-review ├── finishing-a-development-branch ├── writing-plans ├── executing-plans ├── subagent-driven-development ├── writing-skills ├── skill-creator ├── using-git-worktrees └── dispatching-parallel-agents 如何使用 Superpowers 1.]]></description>
</item>
<item>
  <title>OpenCode Skill Creator 使用指南</title>
  <link>https://cfanzp008.github.io/opencode-skill-creator-guide/</link>
  <pubDate>Tue, 07 Apr 2026 14:19:21 &#43;0800</pubDate>
  <author>作者</author>
  <guid>https://cfanzp008.github.io/opencode-skill-creator-guide/</guid>
  <description><![CDATA[OpenCode Skill Creator 使用指南 什么是 Skill Creator Skill Creator 是 Anthropic 官方提供的 Skill 开发助手，帮助开发者创建、优化和打包技能。它是 Claude Code 生态中用于扩展 Agent 能力的重要机制。
在 Claude Code / OpenCode 中，Skill（技能） 是一种可复用的能力扩展包，本质上是一个模块化知识包，可以给 AI 添加：
专业领域知识 固定工作流程 API / 工具使用方式 模板和脚本 简单理解：Skill = 给 AI 写的一份操作说明书
核心价值：
将工作流程固化为可复用的技能 让 AI 助手具备领域专业知识 提高团队协作效率 标准化工作流程 思维导图：Skill Creator 整体架构 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 ┌─────────────────────────────────────────┐ │ Skill Creator 框架 │ │ AI 技能开发与评估工具链 │ └─────────────────────────────────────────┘ │ ┌─────────────────────────────────┼─────────────────────────────────┐ │ │ │ ▼ ▼ ▼ ┌─────────────────────┐ ┌─────────────────────┐ ┌─────────────────────┐ │ 技能结构 │ │ 工作流程 │ │ 核心工具 │ │ (File Structure) │ │ (Workflow) │ │ (Tools) │ └─────────────────────┘ └─────────────────────┘ └─────────────────────┘ │ │ │ ├─SKILL.]]></description>
</item>
<item>
  <title>Superpowers：让 AI Agent 获得专家级能力的技能框架</title>
  <link>https://cfanzp008.github.io/superpowers-opencode-introduction/</link>
  <pubDate>Tue, 07 Apr 2026 11:28:09 &#43;0800</pubDate>
  <author>作者</author>
  <guid>https://cfanzp008.github.io/superpowers-opencode-introduction/</guid>
  <description><![CDATA[Superpowers：让 AI Agent 获得专家级能力的技能框架 背景简介 在 AI 辅助软件开发领域，如何让 AI Agent 像人类专家一样工作一直是核心挑战。传统的 AI 编程工具虽然功能强大，但在处理复杂任务时往往缺乏系统性方法——它们可能会直接开始编码，而不是先理解需求、规划方案。
Superpowers 是由 Keyboardio 联合创始人 Jesse Vincent 开发的一个 AI Agent 技能框架和软件开发方法论。该项目在 GitHub 上已获得超过 134000 个 Star，成为 AI 编程工作流领域的标杆工具。
什么是 Superpowers Superpowers 是一个为 AI 编程 Agent 设计的技能框架和工作流工具集。它的核心理念是：将人类专家的开发习惯和工作流程传授给 AI Agent，让 AI 能够系统性地处理复杂任务，而不是盲目跳入编码。
核心设计理念 Superpowers 基于一个关键洞察：AI Agent 在处理复杂代码时经常失败，但通过特定的技能（Skills）引导，AI 可以表现得像人类专家一样。这些技能封装了：
系统性思考：先理解需求，再制定计划 质量控制：在实现前先验证，在修改后确认 工具使用：正确使用调试、测试、审查等工具 迭代优化：持续改进而非一次完成 技能（Skills）机制 Superpowers 的技能系统受 Simon Willison 提出的 Claude Skills 概念启发。每个技能是一个独立的指令集，可以被 AI Agent 按需加载和使用。技能的核心文件是 SKILL.md，其中包含：
技能的用途说明 使用场景和触发条件 具体的工作流程和步骤 工具映射说明 思维导图：Superpowers 整体架构 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 ┌─────────────────────────────────────────┐ │ Superpowers 框架 │ └─────────────────────────────────────────┘ │ ┌─────────────────────────────────┼─────────────────────────────────┐ │ │ │ ▼ ▼ ▼ ┌─────────────────────┐ ┌─────────────────────┐ ┌─────────────────────┐ │ 核心概念 │ │ 主要技能 │ │ 工具集成 │ │ │ │ │ │ │ └─────────────────────┘ └─────────────────────┘ └─────────────────────┘ │ │ │ ├─技能框架(Skills) ├─brainstorming ├─OpenCode ├─Agent工作流 ├─verification-before-completion ├─Claude Code ├─工具映射 ├─receiving-code-review ├─Codex └─Bootstrap机制 ├─requesting-code-review └─MCP Servers ├─test-driven-development ├─systematic-debugging └─.]]></description>
</item>
<item>
  <title>Tmux &#43; OpenCode 安装使用教程</title>
  <link>https://cfanzp008.github.io/tmux-opencode-tutorial/</link>
  <pubDate>Thu, 02 Apr 2026 14:00:00 &#43;0800</pubDate>
  <author>作者</author>
  <guid>https://cfanzp008.github.io/tmux-opencode-tutorial/</guid>
  <description><![CDATA[Tmux + OpenCode 安装使用教程 背景与简介 在日常开发和运维工作中，我们经常需要同时管理多个终端会话：
远程服务器上运行多个服务 同时查看日志、编辑代码、运行测试 临时断开连接后希望保持任务继续运行 Tmux（Terminal Multiplexer）正是为解决这些痛点而设计的终端复用器。而 OpenCode 作为一个开源的 AI 编程助手，可以大幅提升编码效率。
本文将详细介绍两者的安装与基本使用方法。
什么是 Tmux Tmux 是一个终端复用器，允许在一个终端中运行多个会话、窗口和窗格。它的主要功能包括：
会话管理：保持后台任务运行，即使 SSH 断开也不会中断 分屏功能：将终端分割成多个窗格，同时查看多个任务 会话共享：允许其他人加入到同一个终端会话中 标签式操作：通过快捷键快速切换不同窗口 什么是 OpenCode OpenCode 是一个开源的 AI 编程助手，支持：
代码补全与生成 代码审查与重构 自然语言编程 多模型支持 Tmux 安装 Ubuntu/Debian 1 2 sudo apt-get update sudo apt-get install tmux CentOS/RHEL 1 sudo yum install tmux macOS 1 2 # 使用 Homebrew brew install tmux 验证安装 1 tmux -V Tmux 基本使用 启动与退出 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 # 启动新会话 tmux # 指定会话名称启动 tmux new -s mysession # 退出 tmux（会话保留） Ctrl+b 然后按 d # 列出所有会话 tmux ls # 重新接入会话 tmux attach -t mysession # 杀死会话 tmux kill-session -t mysession 快捷键前缀 Tmux 的所有快捷键都需要先按下 前缀键（默认是 Ctrl+b），然后再按功能键。]]></description>
</item>
<item>
  <title>MiniMax 模型接入 OpenCode 教程</title>
  <link>https://cfanzp008.github.io/minimax-model-opencode-tutorial/</link>
  <pubDate>Thu, 02 Apr 2026 10:00:00 &#43;0800</pubDate>
  <author>作者</author>
  <guid>https://cfanzp008.github.io/minimax-model-opencode-tutorial/</guid>
  <description><![CDATA[MiniMax 模型接入 OpenCode 教程 背景介绍 OpenCode 是一个强大的 AI 编程助手，支持多种大语言模型提供商。MiniMax 作为国内领先的 AI 平台，提供了性能优异的编程模型 M2 系列。本教程将详细介绍如何将 MiniMax 模型接入 OpenCode，实现本地 AI 辅助编程。
什么是 MiniMax MiniMax 是专注于多语言编程能力的 AI 平台，提供以下优势：
强大的代码理解能力：深度理解多种编程语言和框架 长上下文窗口：支持高达 204,800 tokens 的上下文 高速输出：M2.5-highspeed 模型可达 100 tps 输出速度 多模型选择：支持 M2.5、M2.1、M2 等多个版本 支持的模型列表：
模型名称 上下文窗口 输出速度 MiniMax-M2.5 204,800 ~60 tps MiniMax-M2.5-highspeed 204,800 ~100 tps MiniMax-M2.1 204,800 ~60 tps MiniMax-M2.1-highspeed 204,800 ~100 tps 前置条件 在开始配置之前，请确保：
已安装 OpenCode（如果没有安装，可参考官方文档） 已注册 MiniMax 平台账号并获取 API Key 具备基本的命令行操作能力 获取 MiniMax API Key 访问 MiniMax 开放平台 注册并登录账号 在个人中心或开发者设置中获取 API Key 确保账户有足够的 API 调用额度 配置步骤 方法一：环境变量配置 最简单的方式是通过环境变量配置 MiniMax：]]></description>
</item>
<item>
  <title>OpenCode AI 编程助手安装指南</title>
  <link>https://cfanzp008.github.io/opencode-install-guide/</link>
  <pubDate>Wed, 01 Apr 2026 10:00:00 &#43;0800</pubDate>
  <author>作者</author>
  <guid>https://cfanzp008.github.io/opencode-install-guide/</guid>
  <description><![CDATA[OpenCode AI 编程助手安装指南 背景 OpenCode 是一个开源的 AI 编程助手，可以在终端、桌面应用或 IDE 扩展中使用。它支持多种 LLM 提供商，包括 Claude、GPT、Gemini 等，也可以使用免费的内置模型。
OpenCode 拥有超过 12 万 GitHub Star、800 位贡献者和超过 1 万次提交，每月被超过 500 万开发者使用。
本文将详细介绍 OpenCode 在不同操作系统上的安装方法。
系统要求 前置条件 使用 OpenCode 终端版本需要满足以下条件：
终端模拟器：推荐使用现代终端，如：
WezTerm - 跨平台 Alacritty - 跨平台 Ghostty - Linux 和 macOS Kitty - Linux 和 macOS LLM 提供商的 API Key：根据你选择的模型提供商，需要准备相应的 API Key
Linux/macOS 安装 方式一：一键安装脚本（推荐） 这是最简单的安装方式，运行以下命令即可：
1 curl -fsSL https://opencode.ai/install | bash 方式二：使用 Node.js 包管理器 如果你已经安装了 Node.]]></description>
</item>
</channel>
</rss>
