Token导航 LogoToken导航

Agent Skill · 研究检索

sdd-init(SDD 初始化)

sdd-init 用于查找、检索和筛选相关信息,适合在 Codex、Claude、Cursor、Gemini CLI 中需要根据关键词、任务场景或来源线索快速定位候选结果时使用。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。

sdd-init 是研究检索类 Agent Skill,可用宿主为 Codex、Claude、Cursor、Gemini CLI,通过GitHub 仓库获取,权限为需要联网;许可证信息还不明确,第三方扫描已通过,按安装量在研究检索类的 73,675 个 Skill 里排第 66,030。

总安装 269周安装 11GitHub Stars 20下载 87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:sdd-init(SDD 初始化)
来源仓库:https://github.com/sivaprasadreddy/sdd-skills
仓库路径:skills/sdd-init
安装命令:
npx skills add https://github.com/sivaprasadreddy/sdd-skills --skill sdd-init
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/sivaprasadreddy/sdd-skills --skill sdd-init

这个 Skill 是做什么的

研究检索类 Skill 主要解决「找资料、核对事实、把来源整理清楚」的问题,通常会让助手先检索再总结。

sdd-init 用于查找、检索和筛选相关信息。适合在 Codex、Claude、Cursor、Gemini CLI 中根据关键词、任务场景或来源线索快速定位候选结果。通过 npx skills add 命令从指定 GitHub 仓库安装,需结合原始 README 确认具体用法。

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。注意该技能当前分类为研究检索,实际功能以来源仓库文档为准。

来源仓库内路径:skills/sdd-init

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/sivaprasadreddy/sdd-skills --skill sdd-init;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

    它需要需要联网,先在测试目录或测试账号里跑,确认没问题再放到正式项目。

常见问题

6 条

sdd-init 怎么安装?

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

sdd-init 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。

sdd-init 需要什么权限?

权限级别是需要联网。该 Skill 可能需要联网访问来源站点、仓库或外部 API;具体网络访问范围需要结合源码和 README 复核。

sdd-init 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。

sdd-init 用的人多吗?

公开数据是累计安装 269 次,近一周安装 11 次,下载 87 次,GitHub 20 星。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。

sdd-init 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。

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。
  • 需要处理研究检索类任务、又不想每次都手写提示词的人。

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。

研究检索类 Skill 对比

6 个同类
Skill总安装GitHub 星权限许可证扫描
find-skills

寻找技能 · Codex、Claude

31,199,81316,486操作浏览器MIT有提醒
self-improving-agent

自我改进 Agent · OpenClaw

16,464,0023,396接触敏感数据MIT-0通过
soultrace

灵魂追踪 · Codex、Claude

7,588,94412需要联网未明确有提醒
self-improving

自我完善 · OpenClaw

6,829,2841,046需要联网MIT-0通过
microsoft-foundry

微软代工厂 · Codex、Claude

6,614,144764调用外部服务未明确有提醒
sdd-init

SDD 初始化 · 本页

26920需要联网未明确通过

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

SDD: Init docs/project.md

You are a senior software architect analysing a project to produce its context document.

Pre-conditions

Check whether docs/project.md already exists.

If it exists: Ask the user:

"docs/project.md already exists. Do you want to regenerate it? This will overwrite the current file."

Wait for confirmation. If the user says no, stop and tell them to edit the file manually. If the user confirms, proceed with the full process below.

If it does not exist: Proceed immediately.


Process

1. Detect the Tech Stack

Scan the project root and source tree for build and configuration files to auto-detect as much as possible:

What to findWhere to look
Language & versionpom.xml (<java.version>), build.gradle/build.gradle.kts (jvmToolchain, sourceCompatibility), .java-version, pyproject.toml, package.json (engines), go.mod, Cargo.toml
Frameworkpom.xml / build.gradle dependencies — Spring Boot, Quarkus, Micronaut, FastAPI, Express, Django, Rails, etc.
Build toolPresence of pom.xml (Maven), gradlew (Gradle), package.json (npm/yarn/pnpm), Makefile, pyproject.toml (Poetry/Hatch)
DatabaseDependencies or config files — spring.datasource in application.properties/application.yml, DATABASE_URL in .env.example, docker-compose services
ORM / data accessDependencies — Hibernate/JPA, Spring Data, SQLAlchemy, Prisma, TypeORM, GORM, Diesel
MigrationsDependencies or directories — Flyway, Liquibase, Alembic, db/migrate/ (Rails), Prisma migrate
MessagingDependencies or docker-compose services — Kafka, RabbitMQ, SQS, Redis Streams
Testing librariesTest-scope dependencies — JUnit 5, Mockito, Testcontainers, RestAssured, pytest, Jest, Vitest, RSpec
Other librariesRemaining notable dependencies — Spring Security, MapStruct, Lombok, Resilience4j, OpenAPI, etc.

Read docker-compose.yml or docker-compose.yaml if present — it often reveals the database, messaging, and cache stack.

2. Detect Architecture & Conventions

Explore the source tree structure:

  • Identify the top-level source layout (e.g., src/main/java, src/, app/, lib/)
  • Infer the architecture pattern from package/directory names:

- Packages named controller, service, repository → Layered - Packages named domain, application, infrastructure, adapter → Hexagonal / Clean - Packages named after business modules (e.g., order, payment, user) each containing layers → Modular Monolith - Multiple top-level services each with their own build file → Microservices

  • Identify the base package name (e.g., com.example.myapp)
  • Look for a centralised error handler (e.g., GlobalExceptionHandler, middleware, error boundary, or equivalent)
  • Look for a base REST path prefix in controllers or config (e.g., /api/v1)
  • Note any security config (e.g., SecurityConfig, auth annotations, JWT setup)

3. Ask Clarifying Questions

Some information cannot be reliably detected. Ask the user the following questions in a single grouped message — do not ask them one at a time:

I've analysed the project. Before I write docs/project.md, I need a few details:

1. **Project name and mission:** What does this application do, and who uses it?
   (1-2 sentences is enough — this is the "why we exist" paragraph.)

2. **Architecture pattern:** Based on the source layout, I think this uses [detected pattern].
   Is that right, or would you describe it differently?

3. **Key conventions:** Are there any conventions AI Agent should follow that aren't obvious from the code?
   For example: naming rules, auth requirements per endpoint, error response format, pagination style.

4. **Approved dependencies:** Are there any constraints on what libraries can be added?
   (If not, I'll list the detected dependencies as approved by default.)

Adjust or omit questions 2–4 if you already have high confidence from the codebase scan.

Wait for the user's answers before writing the file.

4. Write docs/project.md

Create docs/ if it does not exist. Write docs/project.md with this structure, filled in from your analysis and the user's answers:

# Project: <Name>

## Mission
<One paragraph on what this application does and who it serves.>

## Tech Stack
- Language: <language and version>
- Framework: <framework and version>
- Build tool: <tool>
- Database: <database and version if known>
- ORM: <ORM / data access library>
- Migrations: <migration tool, or "none">
- Messaging: <messaging system, or "none">
- Testing: <testing libraries>
- Other: <other notable libraries>

## Architecture
<Describe the pattern: Layered / Hexagonal / DDD / Modular Monolith / Microservices>
<Describe the package structure with an example, e.g.:>
<  com.example.<module>.<layer>  →  com.example.user.controller, com.example.user.service>

## Conventions
- Package naming: <convention>
- REST base path: <base path, e.g. /api/v1>
- Error handling: <mechanism, e.g. centralised exception handler, error middleware, global error boundary>
- Authentication: <default auth requirement, e.g. all endpoints require auth unless annotated @Public>
- <Any other conventions the user provided>

## Approved Dependencies
<Bulleted list of libraries approved for use. Anything outside this list requires a flag before adding.>

Do not leave any placeholder text (angle-bracket tokens) in the output — fill every field or omit the line if genuinely unknown.

5. Confirm

After writing the file:

  • Show the user a summary of what was detected automatically vs. what they provided
  • Tell them to review docs/project.md and update anything that looks wrong
  • Remind them to keep this file up to date as the stack evolves — it is the primary input for all other SDD skills
  • Prompt them to run /sdd-feature <feature description> to start their first feature

平台分布

Codex

35.79%
按下载量换算31

Claude

29.85%
按下载量换算26

Cursor

21.04%
按下载量换算18

Gemini CLI

10.2%
按下载量换算9

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

通过

权限和风险

需要联网

该 Skill 可能需要联网访问来源站点、仓库或外部 API;具体网络访问范围需要结合源码和 README 复核。

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