Token导航 LogoToken导航

Agent Skill · 研究检索

gherkin(gherkin 搜索)

gherkin 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。

gherkin 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,权限为只读;许可证信息还不明确,第三方扫描有未通过项,按安装量在研究检索类的 73,675 个 Skill 里排第 65,208。

总安装 285周安装 12GitHub Stars 1下载 1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:gherkin(gherkin 搜索)
来源仓库:https://github.com/tsipotu/gherkin-skill
仓库路径:skills/gherkin
安装命令:
npx skills add https://github.com/tsipotu/gherkin-skill --skill gherkin
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/tsipotu/gherkin-skill --skill gherkin

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。

用于查找、检索和筛选相关信息。适合根据关键词、任务场景或来源线索快速定位候选结果。安装命令:npx skills add https://github.com/tsipotu/gherkin-skill --skill gherkin 建议确认权限范围、维护状态及是否会触发联网、命令执行或文件读写。

可结合原始 README 继续核验具体用法。

来源仓库内路径:skills/gherkin

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/tsipotu/gherkin-skill --skill gherkin;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

gherkin 怎么安装?

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

gherkin 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。

gherkin 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。

gherkin 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。

gherkin 用的人多吗?

公开数据是累计安装 285 次,近一周安装 12 次,下载 1 次,GitHub 1 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。

gherkin 安全吗?

第三方扫描有未通过项,风险较高。除非你确认过源码,否则不建议直接安装到常用环境。

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。
  • 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。
  • 第三方扫描存在提醒项或未通过项,需要人工逐条确认。
  • 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。
  • 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。

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
gherkin

gherkin 搜索 · 本页

2851只读未明确未通过

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

Gherkin Scenario Creator

Write Gherkin specifications that capture business behavior clearly and completely. This skill walks users through the process interactively — understanding what they want to specify, thinking through the scenarios they might not have considered, checking for conflicts with existing specs, and producing well-structured .feature files.

How It Works

When a user asks to create Gherkins for a feature, follow this sequence:

1. Understand the Feature

Start by understanding what the user wants to specify. They might say something broad like "create gherkins for the login flow" or something specific like "add a scenario for password reset."

For broad requests, ask before you propose. You'll be tempted to immediately list every scenario you can think of — resist that urge. The user has context you don't: who the actors are, what edge cases matter in their domain, what's already handled elsewhere. Ask 2-3 targeted questions first to shape your thinking. Good questions:

  • Actors: "Who uses this feature? Just regular users, or admins too?" Different actors often mean different scenario sets.
  • State: "What's the starting state? Are there cases where the user is already partially through this flow?"
  • Error handling philosophy: "When something goes wrong (like invalid input), should the system show specific errors or generic messages?"
  • Scope boundaries: "Does this include [related capability], or is that a separate feature?"

Pick the 2-3 most useful questions for this particular feature. Don't ask about things you can reasonably infer. After the user answers, *then* propose your scenario checklist.

Present your suggested scenarios as a checklist and ask which ones they want to include. Add any they suggest that you missed. This collaborative expansion is the core value of the skill — users often think of the happy path and a couple of error cases, but a thorough spec covers edge cases too.

For specific requests, confirm the scope and still consider whether there are closely related scenarios worth including. Mention them briefly — don't force them, just surface them.

2. Scan for Existing Gherkins

Before writing anything, check what already exists in the project.

Search for existing .feature files:

Glob: **/*.feature

Search for Gherkin-style content in other files (some projects keep specs in markdown or docs):

Grep: "Feature:|Scenario:|Given |When |Then " across the project

If existing Gherkin files are found:

  • Read them to understand the project's style and conventions (indentation, tag usage, naming patterns, level of detail)
  • Check whether any existing scenarios already cover the behavior the user is asking for
  • Match the project's existing style — if they use Scenario instead of Example, two-space indentation, specific tag conventions, or particular step phrasing patterns, follow those conventions

If a conflict is detected (an existing scenario covers the same behavior, even if worded differently), flag it clearly:

Heads up: I found an existing scenario in features/catalogue/pricing.feature that already covers this behavior: ``gherkin Scenario: Discount cannot be applied to already-reduced items Given a product marked as "On Sale" When a staff member applies a loyalty discount to it Then they should see "Promotional discounts cannot be combined" `` Your new scenario achieves the same thing. Want to skip it, replace the existing one, or keep both?

Conflict detection is about behavioral equivalence, not identical wording. Two scenarios that test "wrong password shows error" are duplicates even if the steps are phrased differently.

For file placement and naming conventions, read references/organization.md.

3. Write the Scenarios

When writing Gherkin, apply these reference files:

  • Syntax and keywords: references/gherkin-syntax.md — structure, keyword definitions, step arguments, Scenario Outline syntax, and the # language: xx header for non-English scenarios
  • Writing quality: references/best-practices.md — one behavior per scenario, step count, naming, Background, Outline, Rules, and tag guidance
  • Review pass: references/anti-patterns.md — before presenting to the user, scan for technical language in steps, dependent scenarios, vague step names, and mixed Given/When intent

4. Present and Confirm

Show the complete .feature file to the user before writing it. Walk through any design decisions you made:

  • Why you grouped scenarios the way you did
  • Where you used Scenario Outlines vs individual scenarios
  • Any edge cases you included and why

Ask the user to review and suggest changes before finalizing.

5. Write or Append

If an existing .feature file covers this feature area, append the new scenarios to it. Place them logically — under the appropriate Rule block if one exists, or at the end of the file.

If no existing file covers this area, create a new .feature file.

For directory layouts, naming conventions, when to create vs. append, and how to set up structure for brand-new projects, read references/organization.md.

6. Glossary Check

After writing the .feature file, check if the project has a glossary (glossary/glossary.config.yml). If it does:

  1. Read the relevant glossary context files (based on scope) plus core.glossary.yml
  2. Scan the new scenarios for domain terms — nouns in Given/When/Then steps that look like business concepts (e.g., "Cart", "Order", "Wishlist", "Coupon")
  3. Compare against existing glossary terms (checking name, aliases, code_name, and ui_label)
  4. If you find terms that aren't in the glossary, mention them at the end of your summary:
I noticed these terms in the new scenarios that aren't in your glossary yet: - Wishlist — used in 3 scenarios - Gift Card — used in 1 scenario Want to add definitions for any of these?

If the user says yes, hand off to the glossary skill (or walk through the add flow if the glossary skill isn't available).

If the project has no glossary, skip this step silently — don't suggest creating one here.

7. Summary

After writing, provide a brief summary:

  • How many scenarios were added
  • Which file was created or updated
  • Any conflicts that were flagged and how they were resolved
  • Any new terms suggested for the glossary (if applicable)
  • Suggested next scenarios to consider (if the feature has more depth to explore)

Example

User says: "Create gherkins for the password reset flow"

Step 1 — Ask first:

Before I draft the scenarios, a couple of quick questions: 1. Who can reset? Only registered users, or can admins trigger a reset on behalf of someone? 2. Security posture: If someone requests a reset for an email that doesn't exist, should we show a generic "check your email" message (more secure) or tell them the account wasn't found (more helpful)?

Step 1 — After user answers, propose the checklist:

Based on your answers, here's what I'd cover: - User requests a password reset (happy path) - User clicks the reset link and sets a new password - Reset link has expired - Reset link has already been used - User enters mismatched passwords - New password doesn't meet requirements - User requests reset for non-existent email (generic message, per your security preference) Which of these do you want to include? Any others I'm missing?

Step 3 — After confirmation, write the scenarios:

Feature: Password Reset
  Users who forget their password can request a reset link via email
  and use it to set a new password.

  Rule: Requesting a reset

    Scenario: User requests a password reset for a registered email
      Given a registered user with email "morgan@example.com"
      When they request a password reset
      Then they should receive a reset email at "morgan@example.com"
      And they should see "Check your email for reset instructions"

  Rule: Using the reset link

    Scenario: User resets password with a valid link
      Given a valid password reset link for "morgan@example.com"
      When they set a new password "NewSecure123!"
      Then their password should be updated
      And they should be redirected to the login page

    Scenario: User attempts to use an expired reset link
      Given a password reset link that expired 2 hours ago
      When they attempt to set a new password
      Then they should see "This reset link has expired"
      And they should be prompted to request a new one

    Scenario: User attempts to reuse a reset link
      Given a password reset link that has already been used
      When they attempt to set a new password
      Then they should see "This reset link has already been used"

  Rule: Password requirements

    Scenario: User enters mismatched passwords
      Given a valid password reset link
      When they enter "Password1" and confirm with "Password2"
      Then they should see "Passwords do not match"

    Scenario Outline: New password does not meet requirements
      Given a valid password reset link
      When they set a new password "<password>"
      Then they should see "<error>"

      Examples:
        | password | error                                      |
        | short    | Password must be at least 8 characters     |
        | nodigits | Password must contain at least one number  |

平台分布

Codex

35.15%
按下载量换算0

Claude

31.18%
按下载量换算0

Cursor

19.68%
按下载量换算0

Gemini CLI

9.35%
按下载量换算0

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

未通过

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。来源安全扫描存在提醒或未通过项,安装前请自行复核。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