Token导航 LogoToken导航TokenDH.com
研究检索external-servicegithub未标认证来源可访问许可证需确认审计提醒

email-drafting电子邮件起草

Agent Skill

email-drafting 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。

总安装

220

周安装

9

GitHub Stars

106

下载量

71
CodexClaudeCursorGemini CLI

安装说明

本站只整理中文说明和来源信息,不托管安装包,也不代用户安装。

GitHub

来源数

2

许可证

unknown

最后核验

2026-05-01

来源状态

来源可访问

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:email-drafting(电子邮件起草)
来源仓库:https://github.com/mgonto/executive-assistant-skills
仓库路径:skills/email-drafting
安装命令:
npx skills add https://github.com/mgonto/executive-assistant-skills --skill email-drafting
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/mgonto/executive-assistant-skills --skill email-drafting

简介

email-drafting 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。

  • 适用于电子邮件起草、信息整理和线索筛选等研究检索类任务。
  • 通过关键词、任务场景或来源线索进行快速定位和结果筛选。
  • 安装命令:npx skills add https://github.com/mgonto/executive-assistant-skills --skill email-drafting
  • 建议确认权限范围和维护状态,注意是否会触发联网或文件读写操作。

SKILL.md

Email Drafting Skill

Config — read before starting

Read ../config/user.json (resolves to ~/executive-assistant-skills/config/user.json). Extract and use throughout:

  • primary_email, work_email — Gmail accounts
  • scheduling_cc — scheduling assistant email (CC on all scheduling emails, mention in body)
  • scheduling_silent_cc — silent CC for scheduling visibility (do NOT mention in email body)
  • signature — sign-off for all drafts (e.g. "--gonto")
  • name — short name for context

Do not proceed until you have these values.

Debug Logging (MANDATORY)

Read ../config/DEBUG_LOGGING.md for the full convention. Use python3 {user.workspace}/scripts/skill_log.py email-drafting <level> "<message>" ['<details>'] at every key step. Log BEFORE and AFTER every external call (gog gmail, mcporter, todoist-cli). On any error, log the full command and stderr before continuing.

Overview

Auto-draft and manually-requested email drafts for {user.primary_email} and {user.work_email}.

When to Use

  • Gmail hook detects a trigger (intro, scheduling, thanks/ack, positive reply)
  • User asks to draft/reply/send an email
  • Action items cron identifies email follow-ups needed

Architecture

  • Detects triggers and creates Gmail drafts
  • Handles all scheduling (slot-finding, conflict checking, calendar ops)
  • NEVER proposes specific dates/times or creates calendar events

Execution

  • "Reply" always means Reply All — include all original To + CC recipients. Only exclude if user explicitly says to reply to one person. Exception: intro handling moves the introducer to BCC per the intro sequence below.
  • After sending any email/draft, check if it fulfills an open Todoist task (send deck, intro, follow-up, etc.). If yes → complete the task immediately and confirm.
  • Hook/cron triggers: Always run via isolated sub-agent (prevents memory/context contamination)
  • Direct user requests ("draft a reply to X"): Main agent may execute directly, but must still follow all style rules and must NOT read MEMORY.md or daily memory files

Rules (non-negotiable)

Core

  1. Draft-only mode — never send automatically
  2. Mirror inbound language — match the language of the most recent non-automated message in the thread. If the thread has mixed languages, default to the language of the message you're replying to.
  3. Always sign — end every draft with {user.signature}
  4. Low confidence — don't draft; ask user for guidance
  5. No dash punctuation — no em-dash/en-dash in bodies. Use commas/periods.
  6. Humanize — before finalizing any draft, review it against ~/executive-assistant-skills/humanizer/SKILL.md. Remove AI-writing markers: inflated symbolism, promotional tone, em-dash overuse, "delve"/"leverage"/"foster" vocabulary, rule-of-three patterns. Email-specific rules (no dashes, signature, brevity) take precedence over humanizer suggestions if they conflict.

Intro Handling (required sequence)

  1. Thank introducer first
  2. Move introducer to BCC
  3. Reply to introduced contact directly
  4. CC {user.scheduling_cc} and {user.scheduling_silent_cc} for scheduling
  5. Include one line like: "Connecting {user.scheduling_cc_name} to find a time."
  6. Do NOT mention {user.scheduling_silent_cc} in the email body — silent CC only

Scheduling Drafts

  • ALWAYS CC {user.scheduling_cc} AND {user.scheduling_silent_cc}
  • NEVER propose specific dates or times
  • Just confirm willingness to meet + mention scheduling assistant will coordinate
  • Example: "Connecting Alfred to find a time that works"
  • Do NOT mention {user.scheduling_silent_cc} in the email body — she's CC'd silently for visibility

Allowed Auto-Draft Classes

  • Thanks/ack
  • Scheduling intent
  • Positive short replies
  • Intro acceptance

Notification Format

  • account + one-line intent + draft link
  • Example: 📧 Draft ({user.primary_email} → John): intro acceptance. https://mail.google.com/...

Draft Links

After creating a draft via gog, extract the draft ID from the response. Gmail draft URL format: https://mail.google.com/mail/u/?authuser={account_email}#drafts?compose={message_id} Use authuser=m@gon.to or authuser=gonto@hypergrowthpartners.com as appropriate.

Trigger Detection

A) Intro

  • Cues: intro, introduction, meet, connecting you, looping in, cc'ing
  • At least 2 external participants + clear handoff language
  • Apply intro sequence exactly

B) Scheduling Intent

  • Cues: find a time, schedule, availability, next week, calendar
  • Spanish cues: agendar, agenda una, tenés unos minutos
  • CC scheduling contacts, don't propose times

C) Thanks/Ack

  • Cues: thanks, got it, appreciate it, status updates
  • Short acknowledgment + optional one-line next step

D) Positive Short Reply

  • Cues: works for me, sounds good, perfect, great
  • Short affirmative + close

Skip Conditions (do NOT auto-draft)

  • Confidence low / intent ambiguous
  • User already replied in the thread (SENT message exists)
  • Legal, financial, security, hiring-final, sensitive conflict topics
  • Multi-question strategic asks
  • Automated/system/calendar notifications
  • Messages requiring attachments, deep verification, or policy commitments
  • Language unclear or unmirrorable
  • Scheduling confirmations — NEVER auto-draft emails that simply confirm a scheduled time or acknowledge a calendar invite. The scheduling assistant (Alfred/Howie) handles all scheduling coordination. Auto-drafting "confirming our call at X" creates noise and duplicates the scheduler's work. This includes: confirming times proposed by the scheduler, acknowledging calendar invites, and "looking forward to our call" type replies to scheduling threads.

Confidence Gate

Only auto-draft when ALL are true:

  • Trigger class is one of the 4 allowed
  • Language confidently detected and mirrorable
  • Clear recipient intent and next step
  • No skip condition present

Otherwise: ask user.

Drafting Principles

  • Keep it SHORT — drafts are always brief. 2-3 short paragraphs max.
  • No over-explaining — state the point, don't elaborate unless necessary
  • When promising intros: before drafting the intro, search sent emails for previous intros to that person/company, copy the format and tone, and use the same email address
  • When recommending a person/company: use your own words from past emails about them rather than inventing new descriptions
  • Deck/one-pager: say "Hypergrowth Partners deck" (not "our one-pager" or "our deck"). When attaching, frame WHY it's useful (e.g. "where we explain what Hypergrowth is and how we help companies")
  • Future availability: frame as an opportunity, not a brush-off. Position it warmly: "I'd love to reconnect then to explore working together if the timing still makes sense" rather than blunt "let's connect closer to June"
  • Offers of help should use meeting context: read Granola notes from the call and reference specific things discussed. The draft should feel like it came from someone who was in the meeting.
  • No generic "Great meeting today" unless it's explicitly a first meeting (first VC call or first dealflow call).
  • Proposal-first rule: if the commitment is to build/provide a proposal first, do not draft outbound email yet; create TODO only.

Use Grain as primary source for meeting-based drafts

When drafting follow-up emails from meetings, Grain transcript is the primary source (not Granola):

  1. Find the meeting in Grain: mcporter call grain.list_attended_meetings --args '{"limit": 5}'
  2. Fetch transcript: mcporter call grain.fetch_meeting_transcript --args '{"meeting_id": "<id>"}'
  3. Search transcript for email commitments: "I'll send", "I'll email", "I'll share", "let me intro", "I'll follow up", "I'll connect you", etc.
  4. Draft from the transcript — use your actual words and the real conversation context, not Granola's summary.
  5. Fall back to Granola only if Grain has no recording for that meeting.

Style

Read {user.workspace}/style/EMAIL_STYLE.md for the full writing style guide (derived from 200+ real sent emails). Read {user.workspace}/style/FEEDBACK_LOG.md for user corrections — latest overrides win.

Key points:

  • Friendly, concise, action-oriented. Warm but not fluffy.
  • 1–4 short paragraphs, ~6–7 word sentences
  • Context-first openings, straight to point
  • Common opens: "Hey,", "Thanks…", "Perfect…", "Great…"
  • Sign off: {user.signature}
  • No dash punctuation (no em-dash/en-dash)
  • Do: be brief, clear, warm, decisive, include draft link in notification
  • Don't: over-explain, corporate fluff, long formal prose

Templates

  • Primary: {user.workspace}/style/EMAIL_TEMPLATES.md — pattern templates (intros, follow-ups, VC, etc.)
  • Legacy (HGP v1): {user.workspace}/style/email-templates.md — HGP business templates. Use only for HGP-specific business contexts. Primary templates take precedence on conflicts.

Audit Logging (MANDATORY)

After every external action, log it:

  • Draft created: python3 {user.workspace}/scripts/audit_log.py log email_drafted "<recipient>" success '{"account": "<account>", "subject": "<subject>", "type": "<trigger_class>"}'
  • Email sent: python3 {user.workspace}/scripts/audit_log.py log email_sent "<recipient>" success '{"account": "<account>", "subject": "<subject>"}'
  • Draft skipped (low confidence): python3 {user.workspace}/scripts/audit_log.py log email_draft_skipped "<recipient>" skipped '{"reason": "<reason>"}'

Auto-Draft Constraints

  • NEVER create calendar events — only the scheduling assistant handles that
  • Only create email drafts
  • Include --to <sender> explicitly when creating drafts

Auto-Draft WhatsApp Notification (MANDATORY)

Every time a draft is created automatically (via Gmail hook or any automated trigger), you MUST send a WhatsApp notification to {user.whatsapp} with:

✏️ *Auto-draft created*

*To:* <recipient name> (<email>)
*Subject:* <subject>
*Account:* <account>
*Trigger:* <intro/scheduling/thanks/positive reply>

*Draft text:*
> <full draft body — include the complete text so user can review without opening Gmail>

🔗 <Gmail draft link>

Reply "send" to send, or edit in Gmail.

This is non-optional. The user must be able to read and approve the draft from WhatsApp without opening Gmail.

Notification Policy

  • No routine "no change" notifications
  • Alert on: meaningful changes, breakages, time-sensitive items, auto-drafted emails
  • Time-sensitive: approvals, meeting changes, 2FA codes, security, travel changes
  • Evaluate Promotions, suppress Spam/Junk/Trash

适合场景

01

用户想查找某类 Agent Skill 时

02

需要根据任务场景推荐可安装能力包时

03

需要对比不同来源的安装命令和来源信息时

能力概览

能力 1

按任务关键词查找相关 Skills

能力 2

展示可复制的安装命令

能力 3

保留来源站点、仓库和原始说明,方便继续核验

能力 4

展示第三方安全扫描或审计结果

安装后应在对应宿主中按原始 README 的触发条件使用;具体调用方式请以来源页面和 README 为准。

平台分布

Codex

36.39%
按下载量换算26

Claude

28.64%
按下载量换算20

Cursor

18.84%
按下载量换算13

Gemini CLI

9.51%
按下载量换算7

安全审计

Gen Agent Trust Hub

可疑

Socket

通过

Snyk

通过

权限和风险

external-service

该 Skill 可能调用第三方服务、云服务或外部模型 API,使用前需要确认账号、额度、数据发送范围和服务条款。

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 warning/failed 结果,不能写成本站确认安全。当前只有一个来源,正式发布前建议补源仓库或其他目录站核验。

来源信息

继续浏览同类 Skills