Token导航 LogoToken导航

Agent Skill · 研究检索

inbox-reply(收件箱回复)

inbox-reply 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。

inbox-reply 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;许可证信息还不明确,第三方扫描有提醒项,按安装量在研究检索类的 73,675 个 Skill 里排第 57,766。

总安装 424周安装 17GitHub Stars 93下载 137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:inbox-reply(收件箱回复)
来源仓库:https://github.com/extruct-ai/gtm-skills
仓库路径:skills/inbox-reply
安装命令:
npx skills add https://github.com/extruct-ai/gtm-skills --skill inbox-reply
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/extruct-ai/gtm-skills --skill inbox-reply

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。

inbox-reply 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。适用于收件箱内容分析与回复策略相关的信息检索任务。通过 npx skills add 命令从 GitHub 仓库安装,需结合原始 README 确认具体用法。

安装前建议确认权限范围、维护状态及是否触发联网或文件操作。当前无详细功能描述,建议查阅来源仓库获取完整使用说明。

来源仓库内路径:skills/inbox-reply

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/extruct-ai/gtm-skills --skill inbox-reply;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

inbox-reply 怎么安装?

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

inbox-reply 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。

inbox-reply 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。

inbox-reply 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。

inbox-reply 用的人多吗?

公开数据是累计安装 424 次,近一周安装 17 次,下载 137 次,GitHub 93 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。

inbox-reply 安全吗?

第三方扫描有提醒项,说明存在需要人工确认的行为。安装前请打开源码看清它读了什么、联网到哪里。

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。

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。
  • 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。

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
inbox-reply

收件箱回复 · 本页

42493接触敏感数据未明确有提醒

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

Inbox Reply

Manage and reply to lead responses in the email sequencing inbox.

Environment

VariableService
INSTANTLY_API_KEYInstantly

Base URL: https://api.instantly.ai/api/v2

Inputs

InputRequiredSource
Instantly API keyyesenv var
Context fileyescontext file (voice, product, proof points)
Campaign IDnouser selection or list from API

Workflow

Step 1: Fetch unread replies

Fetch unread received emails from the unibox.

List unread: GET /emails with is_unread=true, email_type=received, sort_order=desc, limit=50.

If the user specifies a campaign, add campaign_id filter. If they want focused inbox only, add mode=emode_focused.

Present a summary table:

| # | From | Company | Subject | Preview | Received |

Ask the user which conversations to work on, or process all.

Step 2: Load full threads

For each selected reply, load the full conversation thread.

Get thread: GET /emails?search=thread:{thread_id}&sort_order=asc

This returns the original outbound email(s) and all replies in chronological order, giving full context for drafting a response.

Step 3: Classify reply intent

Read each lead reply and classify into one of these categories:

Interested        → wants to learn more, asks questions, requests a call
Objection         → raises concern but door is not closed (pricing, timing, fit)
Not interested    → clear rejection, asks to stop
Auto-reply        → OOO, bounce, auto-responder
Meeting request   → explicitly asks to book time
Question          → asks a specific question without clear buying signal
Referral          → redirects to someone else at the company

Present classifications to the user for confirmation before drafting.

Step 4: Draft replies

Read the context file for voice rules, product description, proof points, and use case framings. Draft a reply for each conversation based on intent:

Interested       → confirm interest, propose next step (call or resource)
Objection        → address concern directly, offer proof point if relevant
Question         → answer the question, bridge to value
Meeting request  → confirm availability, suggest booking link or times
Referral         → thank them, ask for intro or reach out to referral directly
Not interested   → short, respectful close; do not push
Auto-reply       → skip, no reply needed

Rules for all replies:

  • Match the lead's tone and length — short replies get short responses
  • Reference specifics from their message to show it is not a template
  • Reference the original outbound context when relevant
  • Never start with "I" as the first word
  • No generic openers ("Thanks for getting back to me", "Great to hear from you")
  • One clear call to action per reply, never two
  • No attachments unless the user explicitly provides one
  • Plain text unless the original thread was HTML-formatted

Present each draft alongside the original thread for the user to review and edit.

Step 5: Confirm and send

Before sending, present a final checklist for each reply:

| # | To | Via Account | Subject | Intent | Action |

The user must confirm before any reply is sent.

Send reply: POST /emails/reply with:

  • eaccount: the same email account that sent the original outbound
  • reply_to_uuid: the ID of the lead's most recent message
  • subject: keep the existing thread subject (Re:...)
  • body: the approved reply in {"text": "...", "html": "..."}

Send one at a time. Report success or failure for each.

Step 6: Post-reply housekeeping

After sending:

  1. Mark thread as read: POST /emails/threads/{thread_id}/mark-as-read
  2. Update lead interest status if the classification warrants it: POST /leads/update-interest-status with lead_email and appropriate interest_value
  3. Report final summary:
| # | Lead | Intent | Reply Sent | Status Updated |

Step 7: Skip and defer handling

For replies the user wants to skip or defer:

  • Skip (not interested / auto-reply): mark as read, optionally update interest status
  • Defer: leave unread, note for next session

Never send a reply the user has not explicitly approved.

Reference

See references/instantly-inbox-api.md for full Instantly Unibox and Reply API documentation.

平台分布

Codex

34.25%
按下载量换算47

Claude

30.33%
按下载量换算42

Cursor

16.16%
按下载量换算22

Gemini CLI

9.85%
按下载量换算13

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

可疑

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。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