Token导航 LogoToken导航

Agent Skill · 研究检索

spec_do-next(spec DO Next.js 搜索)

spec_do-next 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。

spec_do-next 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,权限为权限未标注;许可证为 MIT,暂无第三方扫描结果,按安装量在研究检索类的 73,675 个 Skill 里排第 46,860。

总安装 759周安装 31GitHub Stars 公开资料未说明下载 243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:spec_do-next(spec DO Next.js 搜索)
来源仓库:https://github.com/ikatsuba/skills
仓库路径:skills/spec_do-next
安装命令:
npx skills add ikatsuba/skills --skill "spec:do-next"
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。

AgentSkills.tonpx skills
npx skills add ikatsuba/skills --skill "spec:do-next"

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。

针对 Next.js 框架生成具体实现方案。包含路由、样式与状态管理建议。加速 React 生态项目开发节奏。应与实际业务逻辑深度结合。适用宿主包括 Codex、Claude、Cursor、Gemini CLI,接入前应确认版本、权限和运行环境要求。

spec_do-next 属于研究检索类 Skill,可作为该场景下的辅助能力补充。

来源仓库内路径:skills/spec_do-next

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 ikatsuba/skills --skill "spec:do-next";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

    权限没有明确标注,第一次使用时只给一个无关紧要的任务,观察它到底读了哪些文件和网络。

常见问题

6 条

spec_do-next 怎么安装?

在终端执行 npx skills add ikatsuba/skills --skill "spec:do-next"。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

spec_do-next 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_do-next 需要什么权限?

权限级别没有明确标注。第一次使用时建议只给无关紧要的任务,观察它实际读写了哪些内容。

spec_do-next 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。

spec_do-next 用的人多吗?

公开数据是累计安装 759 次,近一周安装 31 次,下载 243 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_do-next 安全吗?

没有查到第三方扫描结果,安全性无法从公开资料判断,安装前请自行审查源码、依赖和命令行为。

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。
  • 许可证明确为 MIT,商用和分发边界好判断。

先想清楚的地方

  • 没有第三方扫描结果,安全性只能靠自己读源码判断。
  • 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

  • 要求有安全扫描结论才肯安装的团队。
  • 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
spec_do-next

spec DO Next.js 搜索 · 本页

759—权限未标注MIT未收录

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

Execute Next Task

Executes the next pending task from a specification's tasks document. This skill finds the first incomplete task and implements it.

When to use

Use this skill when the user needs to:

  • Implement one task at a time with review between tasks
  • Continue work on a partially completed specification
  • Execute tasks incrementally with manual control

Specification Files Structure

All specification documents are located in .specs/<spec-name>/ directory:

FileDescription
.specs/<spec-name>/requirements.mdRequirements and acceptance criteria
.specs/<spec-name>/research.mdResearch findings and chosen solutions
.specs/<spec-name>/design.mdTechnical design and architecture
.specs/<spec-name>/tasks.mdImplementation tasks with checkboxes

Always read all four files to understand the full context before executing tasks.

Instructions

Step 1: Locate and Read Specification Documents

  1. If <args> contains a spec name, look in .specs/<spec-name>/
  2. If no spec name provided, list available specs in .specs/ and use the AskUserQuestion tool to let the user choose
  3. Read and parse all specification documents:

- requirements.md - understand what needs to be built - research.md - understand chosen solutions and their rationale - design.md - understand how it should be built - tasks.md - get the list of tasks to execute

Step 2: Find the Next Task

  1. Scan the document for checkbox markers
  2. Find the first task that is:

- Marked as [-] (in progress) - resume this task first - Or marked as [] (pending) - start this task

  1. Skip tasks marked as [x] (completed)
  2. If all tasks are complete, inform the user

Step 3: Execute the Task

IMPORTANT: Each subtask is executed as a separate subagent and committed independently. Do NOT group subtasks into a single agent or commit.

If the next pending item is a subtask (e.g., 1.2):

  1. Mark subtask as in-progress - Update the subtask checkbox to [-] in tasks.md
  2. Show task info - Display to the user:

- Subtask number and description - Files to create/modify - Requirements being addressed

  1. Launch subagent - Use the Task tool with subagent_type: "general-purpose" to execute the subtask:

- Provide the full subtask description, file paths, and requirements - Include relevant context from the spec (requirements.md, design.md) - Include these rules in the prompt: - Implement directly. Do NOT explore the codebase beyond the files listed in the task. - If you need to understand an existing pattern, read ONLY the specific file — do not launch broad searches. - If tests fail because behavior was intentionally changed, update the tests to match the new behavior. NEVER re-add removed functionality to make old tests pass. - For new fields/entities, ensure they appear in ALL layers: schema, query/mutation, API response type, frontend type, and UI rendering.

  1. Verify result - After the subagent completes:

- Confirm every file listed in the subtask was actually modified (git diff --stat) - If the subtask adds a new field, spot-check it appears in all required layers (schema → query → type → UI) - If verification fails, fix directly or re-run the subagent with specific corrections

  1. Mark subtask as complete - Update the subtask checkbox to [x] in tasks.md only after verification passes
  2. Commit the changes - Use the git:commit skill to commit (see Committing Changes section)
  3. If all subtasks of the parent major task are now complete, mark the major task as [x] in tasks.md and commit this change using the git:commit skill

If the next pending item is a major task with subtasks, start with its first pending subtask using the flow above.

Step 4: Handle Checkpoint Tasks

If the next task is a checkpoint:

  1. Run any verification commands specified
  2. Report the verification results
  3. Mark as complete if all checks pass
  4. Report issues if any checks fail

Step 5: Report Completion

After completing the task:

  1. Summarize what was implemented
  2. Show the next pending task (if any)
  3. Use the AskUserQuestion tool to ask if the user wants to continue, with options like "Continue with next task", "Stop here", "Review changes first"

Committing Changes

After completing each subtask, commit using the git:commit skill:

  1. Stage the changed files related to the subtask
  2. Check if tasks.md is tracked by git (run git check-ignore.specs/<spec-name>/tasks.md). If it is NOT ignored, also stage tasks.md in the same commit so the task progress is captured
  3. Invoke the git:commit skill — it will analyze staged changes, determine the commit type, and create a properly formatted Conventional Commits message

Skip committing if:

  • The user explicitly asked not to commit
  • The subtask only modified the tasks.md file (checkpoint tasks)

Error Handling

  • If the task fails, keep it marked as [-]
  • Report the issue to the user
  • Suggest fixes or ask for guidance

Arguments

  • <args> - The spec name (e.g., "user-auth", "payment-flow")

If not provided, list available specs and ask the user to choose.

平台分布

Claude Code

27.71%
按下载量换算67

Codex

20.36%
按下载量换算49

OpenCode

17.57%
按下载量换算43

Cursor

12.34%
按下载量换算30

windsurf

7.46%
按下载量换算18

kiro-cli

3.33%
按下载量换算8

安全审计

暂无安全审计结果可展示。

权限和风险

权限未标注

当前来源未能明确判断权限范围,默认进入异常复核队列。

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