Token导航 LogoToken导航

Agent Skill · 研究检索

software-engineer(软件工程师)

software-engineer 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。

software-engineer 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,权限为接触敏感数据;许可证为 MIT,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 15,421。

总安装 7,862周安装 315GitHub Stars 11下载 2,545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
MIT

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:software-engineer(软件工程师)
来源仓库:https://github.com/siviter-xyz/dot-agent
仓库路径:skills/software-engineer
安装命令:
npx skills add https://github.com/siviter-xyz/dot-agent --skill software-engineer
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。不同来源提供的安装方式可能略有差异;本站展示可直接复制的安装命令,安装前请核对来源页面。

skills.shnpx skills
npx skills add https://github.com/siviter-xyz/dot-agent --skill software-engineer

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。

software-engineer 用于查找、检索和筛选相关信息,适合在 Codex、Claude、Cursor、Gemini CLI 中快速定位候选结果。它可结合来源仓库、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 https://github.com/siviter-xyz/dot-agent --skill software-engineer 安装。适用于研究检索类任务,需配合宿主环境使用。

来源仓库内路径:skills/software-engineer

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/siviter-xyz/dot-agent --skill software-enginee…;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

    它需要接触敏感数据,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

software-engineer 怎么安装?

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

software-engineer 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。

software-engineer 需要什么权限?

权限级别是接触敏感数据。该 Skill 可能接触密钥、Token、环境变量或敏感配置,应进入高风险复核队列,默认不自动发布。

software-engineer 是免费的吗?

来源标注的许可证是 MIT。本站不代收任何费用,也不托管安装包;能否商用、要不要署名,以许可证原文为准。

software-engineer 用的人多吗?

公开数据是累计安装 7,862 次,近一周安装 315 次,下载 2,545 次,GitHub 11 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。

software-engineer 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。
  • 3 个目录同时收录,说明这个 Skill 流传比较广。
  • 许可证明确为 MIT,商用和分发边界好判断。
  • 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

  • 需要接触敏感数据,装到常用环境前要先评估影响面。
  • 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。
  • 需要许可证清楚、并且有第三方扫描结果的团队项目。

不太适合

  • 环境里有大量密钥、凭证,不方便给第三方 Skill 读取的人。
  • 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
software-engineer

软件工程师 · 本页

7,86211接触敏感数据MIT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

Software Engineering Principles

Core principles and preferences for code style, documentation, and development workflow.

Code Style and Patterns

  • Avoid unnecessary comments: Code should be self-documenting. Reserve comments for non-obvious design decisions, workarounds, or complex logic. Avoid comments that restate what the code obviously does.
  • Clean codebase: Avoid leaving TODO, FIXME, or temporary comments in committed code UNLESS directed. Either implement the feature, create an issue, or remove the comment. Ignore existing ones.
  • Self-documenting code: Prefer clear naming and structure over explanatory comments. Method, class, and member documentation should use language/stack best practices. Don't add useless inline comments next to statements UNLESS they explain confusing or complex behaviour.

Documentation

  • Concise and useful: Documentation should be informative but not verbose. READMEs should focus on essential information without unnecessary elaboration.
  • Structure over verbosity: Prefer well-organized, scannable documentation with clear headings over long paragraphs. Use short examples to illustrate concepts.

Development Workflow

  • Workflow detection: Check if project uses spec-first, TDD, or other structured workflows. Look for:

- docs/ or specs/ directories with specs - Test-first patterns in codebase - Plan files or structured documentation - Follow existing workflow patterns when present

  • No git modifications: Do not use Git commands that modify the repository state (such as git add, git commit, git push) UNLESS directed. Focus on code edits directly. Status and diff commands (git status, git diff) are permitted and encouraged for analysis.
  • Fact-based approach: Do not hallucinate or assume. If you don't know something or need additional context about a framework or technology, search the web or use context7 for up-to-date documentation. If clarification is needed, ask the user before making changes.
  • Constructive disagreement: Do not just accept user direction if a better alternative exists. After reviewing the request, explain your reasoning for why an alternative approach might be better, providing technical justification.
  • Stop and ask: Stop and ask user if:

- Uncertain how to proceed - About to add type ignores, suppressions, or any types - Requirements are unclear - Better approach exists but needs confirmation

  • Backward compatibility: Only consider backward compatibility for public-facing interfaces (APIs, libraries). For greenfield/internal refactoring, unit, integration, & E2E tests serve as confirmation gate unless explicitly directed otherwise.

Code Organization

  • Single responsibility: Components and functions should have a single, clear purpose. Organize code into logical directories with clear separation of concerns.
  • Consistent patterns: Follow established patterns in the codebase. When introducing new patterns, ensure they align with existing architecture and conventions.
  • Automation and efficiency: Prefer automated solutions and efficient workflows. Look for opportunities to reduce manual work and improve developer experience.

Output Formatting

  • No emojis: Do not use emojis in code or output unless explicitly directed
  • Unicode symbols: Unicode symbols (✓, ✗, →, ⚠) are acceptable for user-facing output
  • Color and formatting: Color and formatting encouraged for user-facing output
  • NO_COLOR support: Always respect NO_COLOR environment variable
  • No hardcoded ANSI: Never use hardcoded ANSI color codes - use color libraries (chalk, colors, etc.)

Best Practices

  • Framework conventions: Follow framework and language best practices. Use framework features as intended rather than working around them.
  • Performance awareness: Consider performance implications of code changes, especially for web applications. Prefer static generation and minimal JavaScript when possible.
  • Accessibility: Ensure code is accessible by default. Use semantic HTML, proper ARIA attributes, and test keyboard navigation.

References

For detailed guidance, see:

  • references/workflow-patterns.md - Workflow patterns and practices
  • references/implementation-workflow.md - Unified implementation workflow

平台分布

Claude Code

26.71%
按下载量换算680

Codex

24.17%
按下载量换算615

Gemini CLI

15.7%
按下载量换算400

Antigravity

12.93%
按下载量换算329

OpenCode

7.15%
按下载量换算182

Cursor

3.65%
按下载量换算93

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

通过

权限和风险

接触敏感数据

该 Skill 可能接触密钥、Token、环境变量或敏感配置,应进入高风险复核队列,默认不自动发布。

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。

继续浏览同类 Skills