Token导航 LogoToken导航

Agent Skill · 研究检索

super-plan(超级计划)

super-plan 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。

super-plan 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,第三方扫描已通过,按安装量在研究检索类的 73,675 个 Skill 里排第 61,604。

总安装 349周安装 14GitHub Stars 公开资料未说明下载 113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:super-plan(超级计划)
来源仓库:https://github.com/singh-gur/agent_skills
仓库路径:skills/super-plan
安装命令:
npx skills add https://github.com/singh-gur/agent_skills --skill super-plan
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/singh-gur/agent_skills --skill super-plan

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。

super-plan 提供信息查找、检索与筛选能力,支持按关键词快速定位结果。适用于 Codex、Claude、Cursor、Gemini CLI 等平台的研究与规划类任务。通过 npx skills add 从 GitHub 安装,具体用法请查阅原始项目说明。使用前应评估其是否具备联网、命令执行或文件读写权限。

super-plan 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/super-plan

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/singh-gur/agent_skills --skill super-plan;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

super-plan 怎么安装?

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

super-plan 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。

super-plan 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。

super-plan 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。

super-plan 用的人多吗?

公开数据是累计安装 349 次,近一周安装 14 次,下载 113 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。

super-plan 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
super-plan

超级计划 · 本页

349—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

Super Plan

Create an implementation plan only. Do not implement code while this skill is active unless the user explicitly changes the task.

When to use

Use this skill when:

  • the task is complex enough to benefit from phased execution
  • the work spans multiple files, systems, or decision points
  • the user asks for a plan, roadmap, breakdown, or implementation strategy
  • the change includes refactors, migrations, architecture updates, or risky edits
  • a future executor should be able to complete the work phase by phase with minimal extra context

Do not use this skill for tiny one-step changes unless the user explicitly wants a written plan.

Core behavior

  • Plan only. Do not silently switch from planning to implementation.
  • Explore before planning. Base the plan on the actual repository, not guesses.
  • Ask focused clarifying questions when requirements, constraints, or priorities are ambiguous.
  • Break work into balanced phases that are meaningful, reviewable, and independently executable.
  • Prefer plans that can survive handoff: each phase should contain enough context for another agent or human to execute it.
  • Keep the skill harness-agnostic. Refer to capabilities generically, not by product-specific tool names.
  • After writing the plan, summarize it for the user without pasting the entire file back into chat unless asked.

Planning workflow

  1. Understand the request

- Restate the planning goal internally. - Identify unknowns, constraints, and success criteria. - If a missing decision materially changes the plan, ask the user before finalizing it.

  1. Explore the codebase

- Inspect relevant files, structure, conventions, dependencies, and existing patterns. - Look for architecture boundaries, integration points, risk areas, and any related tests or docs. - Prefer evidence from the repository over assumptions.

  1. Detect repository context

- Check whether the project appears to use version control. - If version control is relevant to execution planning, ask the user which checkpointing strategy they prefer. - Keep the wording generic and workflow-oriented rather than tied to a specific harness.

  1. Design balanced phases

- Each phase should usually represent roughly 30-90 minutes of focused work. - A phase must be self-contained, have clear prerequisites, define outputs, and include verification criteria. - Avoid micro-phases that create coordination overhead without producing a meaningful checkpoint. - Split oversized phases that mix unrelated outcomes or become difficult to review. - Mark phases that can run in parallel.

  1. Choose the plan destination

- Ask the user where to save the plan before writing it. - Offer: - PLAN.md in the repository root - plans/<task-name>.md for organized multi-plan workflows - If using the plans/ option, generate a kebab-case filename from the task and let the user adjust it.

  1. Write exactly one plan file

- The plan file is the main deliverable. - Keep it structured, concrete, and updateable during execution.

  1. Summarize for the user

- Give a concise recap of the overall approach, main phases, major risks or open questions, and saved file path.

Plan requirements

Structure the plan so it is easy to execute and maintain during implementation.

Recommended sections

# Implementation Plan: <Task Name>

## Overview
## Global Context
## Architecture Decisions
## Assumptions
## Phase Strategy
## Phases
## Phase Dependencies
## Risks
## Questions for User

Phase template

Each phase should include:

  • Objective
  • Status: Not Started
  • Complexity
  • Estimated Time
  • Prerequisites
  • Context for this Phase
  • Files table
  • Implementation Tasks as markdown checkboxes
  • Execution Tracking Rules
  • Verification
  • Completion Gate
  • Outputs

Execution tracking rules

Write phases so the plan can be updated during implementation:

  • Initial status must be Not Started.
  • When execution begins, the phase status should become In Progress before phase-specific workflow setup.
  • Completed tasks should be checked off.
  • Skipped or superseded tasks should be struck through rather than falsely marked done.
  • A phase should become Complete only after the user reviews the result and explicitly confirms it is done.
  • If the project uses a checkpointing workflow, cleanup/finalization for that phase should happen immediately after user-confirmed completion.

Version-control guidance

If the repository uses version control and the user wants workflow guidance, include one strategy section matching their choice. Keep the terminology universal and describe the workflow by intent.

Possible strategies:

  • separate working directories for each phase
  • separate feature branches for each phase
  • checkpoint tags on a shared branch
  • decide the strategy separately when each phase starts

For the chosen strategy, explain:

  • how execution starts for a phase
  • how dependencies affect later phases
  • what must happen when the user confirms the phase is complete
  • how to avoid leaving unfinished phase state behind

Do not depend on harness-specific commands or tool names in the skill text.

Quality bar

A good plan produced by this skill should:

  • reflect real repository findings
  • surface important assumptions and risks
  • provide enough context for handoff to another executor
  • keep sibling phases reasonably balanced
  • give objective verification criteria where possible
  • avoid vague steps such as "update code as needed"

Important rules

  • Do not implement code while acting in planning mode.
  • Do not invent repository details you have not inspected.
  • Ask at most the focused questions needed to remove meaningful ambiguity.
  • Keep plan instructions executable across different agentic environments.
  • Do not mention product-specific tool names unless the user explicitly asks for a harness-specific variant.
  • After saving the plan, respond with a concise summary and the saved path.

平台分布

Codex

33.95%
按下载量换算38

Claude

29.85%
按下载量换算34

Cursor

19.93%
按下载量换算23

Gemini CLI

8.79%
按下载量换算10

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