Token导航 LogoToken导航

Agent Skill · 研究检索

briefing(briefing 效率)

briefing 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。

briefing 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 53,604。

总安装 528周安装 22GitHub Stars 公开资料未说明下载 176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:briefing(briefing 效率)
来源仓库:https://github.com/sjunepark/custom-skills
仓库路径:skills/briefing
安装命令:
npx skills add https://github.com/sjunepark/custom-skills --skill briefing
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/sjunepark/custom-skills --skill briefing

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。

briefing 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。它支持基于关键词、任务场景或来源线索进行信息匹配,适用于研究检索类需求。

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。安装前建议核实权限范围、维护状态,以及是否涉及联网、命令执行或文件读写操作。该技能适用于需要高效信息聚合的场景,但需人工核验其实际行为与项目需求是否匹配。

来源仓库内路径:skills/briefing

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/sjunepark/custom-skills --skill briefing;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

briefing 怎么安装?

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

briefing 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。

briefing 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。

briefing 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。

briefing 用的人多吗?

公开数据是累计安装 528 次,近一周安装 22 次,下载 176 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。

briefing 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
briefing

briefing 效率 · 本页

528—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

Briefing

Give the user a short, decision-ready view of the current task. Restore their task-scoped working context without dumping the whole conversation or codebase back at them.

The user is the decision-maker. Your job is to surface the relevant state, code, and constraints clearly enough that they can steer the next step.

Core Job

Restore the user's working context for the current task so they can make the next decision.

This skill is not for broad repository explanation. It is for reconnecting the user to the few details that matter now.

Core Behavior

  1. Work from the current request first.
  • Treat the current task or question as the scope boundary.
  • Read the current conversation before reading code.
  • Pull in only the files and docs needed to explain the current state.
  1. Stay task-scoped.
  • Focus on the current request or session, not the full project history.
  • Include broader roadmap or open questions only when they materially affect the current task.
  • Avoid generic repository summaries.
  1. Keep the briefing brief.
  • Default to 8-14 bullets total, or the equivalent in short sections.
  • Prefer high-signal facts over exhaustive detail.
  • Do not walk line by line through code unless the user explicitly asks.
  1. Be mostly descriptive.
  • Explain the current state, relevant code, and constraints.
  • Light recommendations are allowed when they help orientation, such as the likely next step or the main thing to watch.
  • Do not turn the briefing into a plan review, code review, or design critique unless the user asks for that mode.

Sources

Read sources in this order and stop as soon as the current task can be explained confidently:

  • the current conversation
  • the smallest set of relevant files in the codebase
  • relevant local docs such as AGENTS.md, ARCHITECTURE.md, or nearby module docs when they change how the task should be understood

Use file paths when they materially help the user reconnect to the code.

Briefing Styles

Default to a status briefing.

Do not depend on markdown headings or a rigid internal mode switch to decide behavior. Instead, infer the user's emphasis from their explicit request:

  • If they ask to catch up, brief current state.
  • If they ask what you are likely to do, emphasize likely implementation approach.
  • If they ask to teach the codebase, emphasize roles, flows, and file relationships.
  • If they ask for help deciding, emphasize constraints, tradeoffs, and the facts that matter now.

Treat these as variations of one job: produce a concise task-scoped briefing. If the request is mixed, combine the needed emphases without turning the response into a long menu.

Non-Goals

Do not use this skill to produce:

  • a full architecture summary for the repository
  • a code review or design review
  • a project roadmap reconstruction unless the current task depends on it
  • a verbose recap of the whole conversation
  • a line-by-line walkthrough of implementation details unless the user explicitly asks

Workflow

  1. Identify the user's requested emphasis.
  • If the user explicitly asks for current state, use a status emphasis.
  • If they ask what you will likely do next, use a plan emphasis.
  • If they ask to understand the relevant code, use a teach emphasis.
  • If they ask what matters for a decision, use a decision emphasis.
  • If the request is ambiguous, default to status.
  1. Recover the task boundary.
  • State the current request in concrete terms.
  • Distinguish the present task from earlier side discussions if needed.
  1. Gather only the relevant context.
  • Start with the current conversation.
  • Read the smallest set of files and docs that explain the current state.
  • Prefer entry points, touched files, nearby interfaces, and stable docs over broad exploration.
  1. Synthesize for fast orientation.
  • Collapse raw findings into the few facts the user needs now.
  • Name the relevant files and modules directly.
  • Explain why each one matters to the current task.
  1. Trim aggressively.
  • Remove repeated points.
  • Remove code trivia and low-signal history.
  • Remove speculative statements unless they are clearly labeled.

Output

Default to short sections with flat bullets.

Use this shape unless the user asks otherwise:

Current Task

  • One to three bullets on what is being worked on.

Relevant Code

  • The key files or modules, with a short note on why each matters.

What Matters Now

  • The most important implementation facts, behavior, or constraints for this task.

Decision Context

  • Only the facts the user needs to make the next decision.
  • Include a light next-step note only when it helps.

Adjust emphasis based on the user's request:

  • status: center the current implementation state.
  • plan: center the likely approach and touched areas.
  • teach: center roles, boundaries, and flow.
  • decision: center constraints, tradeoffs, and high-impact unknowns.

Communication Rules

  • Prefer the smallest useful briefing over a comprehensive one.
  • Prefer concrete file references over abstract architecture language.
  • Prefer "what matters now" over "everything that exists."
  • Keep explanations compact but complete enough for decision-making.
  • Do not repeat the conversation back to the user.
  • Do not pad the output with generic advice.
  • Do not create open questions unless the current task genuinely centers on them.
  • If evidence is partial, say what is known and what is inferred.

Example Triggers

  • "Use $briefing and catch me up on what matters for this task."
  • "Before you implement this, give me a briefing on the relevant code."
  • "Brief me on the current state of this feature."
  • "Teach me the part of the codebase related to this request."
  • "Give me a decision briefing for this change."

平台分布

Codex

33.98%
按下载量换算60

Claude

29.59%
按下载量换算52

Cursor

20.85%
按下载量换算37

Gemini CLI

10.45%
按下载量换算18

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