做什么
将AI协助的产品交付工作流程标准化,将新想法或混乱的储存库变成一个有条理的建设循环,包括`设计-文件.md'、`tech-stack.md'、`实施-计划.md'和持续的`记忆-银行/'。 用于Codex或Claude Code,当拖动绿地工作、将存储器正常化为文件第一进程、或在执行`进步.md'和`architecture.md'保持当前状态的同时执行一个经过验证的步骤时.
技能库 智客分类:文档办公 vibe-coding-standard-workflow
将AI协助的产品交付工作流程标准化,将新想法或混乱的储存库变成一个有条理的建设循环,包括`设计-文件.md'、`tech-stack.md'、`实施-计划.md'和持续的`记忆-银行/'。 用于Codex或Claude Code,当拖动绿地工作、将存储器正常化为文件第一进程、或在执行`进步.md'和`architecture.md'保持当前状态的同时执行一个经过验证的步骤时.
官方网址:skills.sh
先看中文介绍;官方 description 原文单独保留,不改写 SKILL.md。
将AI协助的产品交付工作流程标准化,将新想法或混乱的储存库变成一个有条理的建设循环,包括`设计-文件.md'、`tech-stack.md'、`实施-计划.md'和持续的`记忆-银行/'。 用于Codex或Claude Code,当拖动绿地工作、将存储器正常化为文件第一进程、或在执行`进步.md'和`architecture.md'保持当前状态的同时执行一个经过验证的步骤时.
官方 description 未单独写出 Use when。按规范,代理会在用户任务与这段 description 的关键词匹配时激活本技能。
按 Agent Skills 渐进披露:启动时只加载 name 与 description(约 100 token);任务匹配后才读入整份 SKILL.md 正文;scripts/、references/、assets/ 仅在需要时再读。 本文件正文结构:Vibe Coding Standard Workflow、Target Artifacts、Bootstrap Workflow、1. Normalize the initial idea、2. Convert the PRD into a real design document、3. Recommend the technology stack。 其中含规范建议的小节:分步指令。
文件分析:除 SKILL.md 外,正文引用了 references/codex.md、references/claude-code.md、references/prompts.md,属于带资源的技能包,这些文件按需再读。
Standardized AI-assisted product delivery workflow for turning a new idea or messy repository into a disciplined build loop with `design-document.md`, `tech-stack.md`, `implementation-plan.md`, and a persistent `memory-bank/`. Use in Codex or Claude Code when bootstrapping greenfield work, normalizing a repository into a document-first process, or executing implementation one validated step at a time with `progress.md` and `architecture.md` kept current.
Vibe Coding Standard WorkflowTarget ArtifactsBootstrap Workflow1. Normalize the initial idea2. Convert the PRD into a real design document3. Recommend the technology stack4. Bootstrap project memory and agent instructions重要提示5. Produce the implementation plan6. Refine the plan until it is executableExecution Loop7. Execute exactly one plan step at a time
来源分类:skills.sh agent-skill
namevibe-coding-standard-workflowdescriptionreferences/codex.mdreferences/claude-code.mdreferences/prompts.md以下路径提取自原文;文件是否齐全请以来源仓库中的完整目录为准。
具体调用语法与可用工具以目标 Agent 客户端为准。 查看调用机制说明 ↗
先选择目标 Agent 和安装范围,保留技能包的附属文件,安装后检查客户端能否发现该技能。
该技能引用了附属文件,请从来源获取完整目录;仅复制 SKILL.md 可能缺少依赖。
复制安装指令给支持 Agent Skills 的代理,确认其中的目标目录与客户端匹配。
把 Agent Skill「vibe-coding-standard-workflow」安装到我的项目:SKILL.md 原文与官方 description 见 https://zicq.com/zh/skills/skl-3da7fddfd4dcc31d-Vibe-%E7%BC%96%E7%A0%81%E6%A0%87%E5%87%86%E5%B7%A5%E4%BD%9C%E6%B5%81%E7%A8%8B.html 请存为 .cursor/skills/vibe-coding-standard-workflow/SKILL.md 或 .claude/skills/vibe-coding-standard-workflow/SKILL.md,frontmatter 的 name 与 description 保持原样,不要改写。 该技能还带 scripts/、references/、assets/ 等文件,请从 https://github.com/woodyxu/vibe-coding-standard-workflow 取完整目录,不要只建一个 SKILL.md。
需要 Node.js 与 npx。先查看仓库技能列表,确认实际名称。
npx skills add 'https://github.com/woodyxu/vibe-coding-standard-workflow' --list
npx skills add 'https://github.com/woodyxu/vibe-coding-standard-workflow' --skill 'vibe-coding-standard-workflow'
CLI 会交互选择目标 Agent,默认安装到项目;用户级安装使用 -g。先通过查看命令核对仓库内容,再用 npx skills list 检查已安装技能。
design-document.md, tech-stack.md, implementation-plan.md, and a persistent memory-bank/. Use in Codex or Claude Code when bootstrapping greenfield work, normalizing a repository into a document-first process, or executing implementation one validated step at a time with progress.md and architecture.md kept current.Turn a rough idea into a controlled delivery loop. Use this skill to force document-first planning, explicit clarification, small testable implementation steps, and durable project memory for later handoffs.
Create and maintain these files:
PRD.md as the temporary idea capture filememory-bank/design-document.mdmemory-bank/tech-stack.mdmemory-bank/implementation-plan.mdmemory-bank/progress.mdmemory-bank/architecture.mdAGENTS.md and/or CLAUDE.md, depending on the agent environmentTreat memory-bank/ as the long-lived source of truth after the bootstrap phase.
Start by producing a short PRD.md that captures:
Keep it lightweight. The purpose is to create enough structure for deeper questioning, not to finish the specification.
Read PRD.md, then ask targeted clarification questions before writing design-document.md. Do not jump directly into the design document if the PRD is still ambiguous.
The resulting design document should make implementation decisions possible. Cover at least:
If the user answers questions incrementally, keep refining the document until the design is executable.
Derive tech-stack.md from design-document.md, not from guesses. Optimize for simple and robust over trendy and broad.
For each major layer, choose the minimum viable stack:
Keep rationale short but explicit.
After the design and stack are stable:
PRD.md if the workflow or user instruction allows itAGENTS.md and/or CLAUDE.md include the required memory-bank rulesmemory-bank/design-document.md, tech-stack.md, and implementation-plan.md into memory-bank/memory-bank/progress.md and memory-bank/architecture.mdIf only one of AGENTS.md or CLAUDE.md exists, update that file. If both exist, keep them aligned.
Append this block:
## 重要提示
- 写任何代码前必须完整阅读 memory-bank/@architecture.md
- 写任何代码前必须完整阅读 memory-bank/@design-document.md
- 每完成一个重大功能或里程碑后,必须更新 memory-bank/@architecture.md
Generate implementation-plan.md from design-document.md and tech-stack.md.
Every step in the plan must be:
Each step must include its own validation instruction or test expectation. Do not allow vague steps like "build the backend" or "finish the frontend."
Read the full memory-bank/ and ask whatever questions are still necessary to make the plan unambiguous. Continue until the implementation plan is clear enough that another agent could execute it without re-discovering product intent.
Update the documents after every clarification round.
Before writing code, read:
memory-bank/architecture.mdmemory-bank/design-document.mdmemory-bank/implementation-plan.mdmemory-bank/progress.mdThen execute only the current plan step.
Assume the user is the test gate unless they explicitly delegate test execution. Do not start the next step until the current one is verified.
After the user confirms the step:
memory-bank/progress.mdmemory-bank/architecture.md with new or changed files and why they existFor every later step, reload the memory-bank/ files and the latest progress.md before continuing. If context quality has degraded, explicitly clear and rebuild context instead of relying on memory.
implementation-plan.md free of code.N+1 before step N is validated.progress.md as the execution log, not as a design document.architecture.md as a current map of important files and responsibilities.文档办公
为结构化的代理内存和可堆肥技能所打入的知识图. 在创建/征服实体(Person, project, Task, Evention, Document)时使用,链接相关对象,强制约束,规划多步动作作为图变,或技能需要共享状态时使用. 触发到"记住","我知道什么","链接X到Y",…
文档办公
使用纳米-pdf CLI编辑带有自然语言指令的PDF.
文档办公
创建,检查,并编辑有可靠样式的Microsoft Word文档和DOCX文件,编号,跟踪更改,表格,章节,并进行相容性检查. 当 (1) 任务涉及 Word 或 ".docx " 时使用; (2) 文件包括跟踪的更改,评论,字段,表格,模板,或页面布局限制; (3) 文档必须在不…
文档办公
通过 Maton 网关调用第三方 API, 它为用户已经连接的应用程序注入证书 。 当用户命名一个连接的应用程序和其中的具体行动时使用此技能——读取一个邮箱,查询一个CMM,归档一个问题,更新一个电子表格,通过一个连接的搜索或刮取提供者运行一个查询. 每一次通话都到用户连接的应用…