Token导航 LogoToken导航

Agent Skill · 开发

spec-driven-apply(规格驱动应用)

spec-driven-apply 用于处理 GitHub 仓库、Issue、Pull Request 和代码协作信息,适合在 Codex、Claude、Cursor、Gemini CLI 中需要围绕仓库状态、代码变更或协作事项进行整理时使用。可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。

spec-driven-apply 是开发类 Agent Skill,可用宿主为 Codex、Claude、Cursor、Gemini CLI,通过GitHub 仓库获取,权限为只读;许可证信息还不明确,第三方扫描已通过,按安装量在开发类的 21,746 个 Skill 里排第 13,798。

总安装 823周安装 35GitHub Stars 公开资料未说明下载 288核验 2026-05-01

安装说明

本站只整理中文说明,不托管安装包,也不代用户安装。
GitHub
许可证
unknown

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:spec-driven-apply(规格驱动应用)
来源仓库:https://github.com/kw12121212/auto-spec-driven
仓库路径:skills/spec-driven-apply
安装命令:
npx skills add https://github.com/kw12121212/auto-spec-driven --skill spec-driven-apply
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 npx skills 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

skills.shnpx skills
npx skills add https://github.com/kw12121212/auto-spec-driven --skill spec-driven-apply

这个 Skill 是做什么的

开发类 Skill 面向写代码、改代码和排查问题,一般会读写仓库文件,部分还会执行构建或测试命令。

用于处理 GitHub 仓库、Issue、Pull Request 和代码协作信息,适合在 Codex、Claude、Cursor、Gemini CLI 中围绕仓库状态、代码变更或协作事项进行整理。可结合来源仓库、安装命令和原始 README 继续核验具体用法。

安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。通过 npx skills add 命令从指定 GitHub 仓库安装 spec-driven-apply 技能。当前无额外说明,需参考原始 SKILL.md 获取完整功能细节。

来源仓库内路径:skills/spec-driven-apply

上手步骤

4 步
  1. 1

    确认自己的宿主工具

    公开资料标注支持 Codex、Claude、Cursor、Gemini CLI;先确认自己用的是其中之一,再按对应工具的 Skills 目录放置。

  2. 2

    拿到安装命令

    可以直接复制的命令是 npx skills add https://github.com/kw12121212/auto-spec-driven --skill spec-drive…;命令里的占位符换成自己的路径或密钥再执行。

  3. 3

    安装后重启宿主

    命令执行完要完全退出宿主工具再打开,让新装的 Skill 被重新扫描到,否则列表里可能看不到。

  4. 4

    按最小权限试用

    它标注为只读,先用只读任务试一次,确认行为符合预期再扩大使用范围。

常见问题

6 条

spec-driven-apply 怎么安装?

在终端执行 npx skills add https://github.com/kw12121212/auto-spec-driven --skill spec-driven-apply。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

spec-driven-apply 支持哪些 AI 工具?

公开资料标注支持 Codex、Claude、Cursor、Gemini CLI。其他工具只要能读取本地 Skill 目录,一般也能手动放进去,但要自己确认触发方式是否一致。

spec-driven-apply 需要什么权限?

权限级别是只读。该 Skill 主要提供规则、说明或参考内容,本身偏只读;真正读写文件、联网或执行命令仍取决于宿主 Agent 的任务。

spec-driven-apply 是免费的吗?

公开资料没有标注明确许可证,收费和商用限制都无法确认。正式使用前请到来源仓库核对授权说明。

spec-driven-apply 用的人多吗?

公开数据是累计安装 823 次,近一周安装 35 次,下载 288 次。这些数字来自第三方目录统计,只能当作热度参考,不代表质量或安全性。

spec-driven-apply 安全吗?

第三方扫描结果为通过,但这只覆盖已知的静态检查项,不能替代自己读一遍源码和安装命令。

值得选的地方和要先想清楚的地方

值得选的地方

  • 同时适配 4 种工具,换客户端不用重新找替代品。
  • 第三方扫描已通过,常见的高风险行为没有命中。
  • 标注为只读权限,试用阶段不容易误改本地文件。

先想清楚的地方

  • 许可证没有明确标注,能不能商用、能不能再分发都不确定。
  • 第三方目录整理,版本更新和行为细节以来源仓库为准。

适不适合你

更适合

  • 已经在用 Codex,想直接扩展它能力的人。
  • 需要处理开发类任务、又不想每次都手写提示词的人。
  • 只想先试试效果、不希望它改动本地文件的人。

不太适合

  • 需要明确授权才能用到公司项目里的场景。
  • 需要厂商级技术支持和 SLA 的生产系统。

开发类 Skill 对比

6 个同类
Skill总安装GitHub 星权限许可证扫描
skill-vetter

技能审查员 · OpenClaw

8,779,018986操作浏览器MIT-0通过
ontology

本体论 · OpenClaw

6,636,960576只读MIT-0通过
entra-app-registration

内部应用程序注册 · Codex、Claude

6,631,872764接触敏感数据未明确通过
entra-app-registration

内部应用程序注册 · Codex、Claude

3,757,680201接触敏感数据未明确通过
nano-banana-pro

纳米香蕉专业版 · OpenClaw

3,526,898364接触敏感数据MIT-0有提醒
spec-driven-apply

规格驱动应用 · 本页

823—只读未明确通过

按安装量取同分类前几名;数字来自第三方目录统计,只作热度参考,不代表质量结论。

SKILL.md 原文

You are helping the user implement a spec-driven change.

This Skill's Commands

If you cannot remember the exact command used by this skill, look it up here before running anything. Do not guess.

modify: node {{SKILL_DIR}}/scripts/spec-driven.js modify
apply: node {{SKILL_DIR}}/scripts/spec-driven.js apply <name>

Prerequisites

The .spec-driven/ directory must exist at the project root. Before proceeding, verify:

ls .spec-driven/

If this fails, the project is not initialized. Run /spec-driven-init first.

Steps

  1. Select the change — run node {{SKILL_DIR}}/scripts/spec-driven.js modify to list active changes. Ask which change to apply. If already specified, use it.
  2. Load context — read all artifacts:

- You MUST treat all prior conversational context as stale, unreliable, and non-authoritative. - You MUST NOT use prior chat context as a source of truth for requirements, task state, implementation details, or completion status. - You MUST rebuild working context from the current change artifacts, relevant base specs, and the current repository state before taking any implementation action. - If prior chat context differs from the files or repository state in any way, you MUST discard the prior chat context and follow the files and repository state only. - Do not use prior chat context unless it has been explicitly re-validated against the current files and repository state. - .spec-driven/changes/<name>/proposal.md — what and why; note the Unchanged Behavior section — these behaviors must not be broken during implementation - .spec-driven/changes/<name>/specs/ — delta spec files (mirror of main specs/ structure) - .spec-driven/changes/<name>/design.md — approach and decisions - .spec-driven/changes/<name>/tasks.md — the checklist - .spec-driven/changes/<name>/questions.md — open and resolved questions Also read: - .spec-driven/config.yaml — project context; treat rules as binding constraints; if fileMatch entries are present, apply matching rules when editing files whose paths match the glob pattern - .spec-driven/specs/INDEX.md — identifies all existing spec files - Every spec file in INDEX.md that is relevant to this change — read full content to understand current requirements before writing code - Mapping frontmatter in relevant main and delta spec files; use mapping.implementation and mapping.tests as starting context for files to inspect

  1. Check task status — run: node {{SKILL_DIR}}/scripts/spec-driven.js apply <name> Show the user the task summary (total, complete, remaining).
  2. Gate on open questions — read questions.md and scan for - [] Q: entries:

- If any open questions are found, list each one and stop before implementation - For each open question, present a structured user-facing block with: - Question - Explanation - Impact - Recommendation - Explanation must clarify why the question is still unresolved after reading the change artifacts, relevant main specs, and current repository state - Impact must describe what implementation choice, task, behavior, or unchanged-behavior guarantee depends on the answer - Recommendation may suggest a preferred answer, but only as a suggestion - Do not implement tasks, edit implementation files, or mark the question resolved until the user explicitly answers or confirms the decision - Only continue once all questions are resolved in questions.md under ## Resolved with an A: answer

  1. Implement tasks — work through each - [] task in order:

- Read relevant code before making changes - Implement the task - Verify the change does not violate any Unchanged Behavior listed in proposal.md - For ## Testing tasks: actually run the tests (lint, unit tests) and confirm they pass before marking complete - Mark it complete in tasks.md by changing - [] to - [x] - Briefly confirm what was done before moving to the next task

  1. Update delta specs — after all tasks are done, re-read each file in changes/<name>/specs/ and verify it accurately reflects what was actually implemented:

- If the implementation diverged from the original plan, update the affected files - If additional spec files need to be created or modified, do so now - Update delta spec mapping frontmatter so mapping.implementation and mapping.tests list the actual implementation and test files changed or relied on

  1. On completion — when all tasks are done and delta spec is accurate:

- Run node {{SKILL_DIR}}/scripts/spec-driven.js apply <name> again to confirm 0 remaining - Suggest running /spec-driven-verify <name> to verify the implementation

Rules

  • Read existing code before modifying it
  • Complete tasks in order unless there is a clear dependency reason to skip
  • Mark tasks complete immediately after implementing them, not in bulk at the end
  • If a task is ambiguous, read proposal.md and design.md before asking the user
  • Do not implement tasks that are already marked - [x]
  • Open questions require explicit user resolution before implementation starts
  • Recommendations do not count as resolved answers until the user confirms them
  • Keep implementation and test mappings in frontmatter, not in requirement prose

平台分布

Codex

38.25%
按下载量换算110

Claude

27.43%
按下载量换算79

Cursor

16.65%
按下载量换算48

Gemini CLI

9.83%
按下载量换算28

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

通过

权限和风险

只读

该 Skill 主要提供规则、说明或参考内容,本身偏只读;真正读写文件、联网或执行命令仍取决于宿主 Agent 的任务。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。目前只有一个公开来源,建议同时核对源仓库。

继续浏览同类 Skills