Token导航 LogoToken导航

Agent Skill · 研究检索

message(message 效率)

message 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。

message 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,权限为操作浏览器;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 57,350。

总安装 436周安装 18GitHub Stars 51下载 143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:message(message 效率)
来源仓库:https://github.com/henkisdabro/wookstar-claude-plugins
仓库路径:skills/message
安装命令:
npx skills add https://github.com/henkisdabro/wookstar-claude-plugins --skill message
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/henkisdabro/wookstar-claude-plugins --skill message

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。

message 用于查找、检索和筛选相关信息。适合在 Codex、Claude、Cursor、Gemini CLI 中根据关键词快速定位候选结果。通过 npx skills add 命令从指定 GitHub 仓库安装,需确认权限和维护状态。使用前建议核验具体用法,注意是否会触发联网、命令执行或文件读写操作。

涉及敏感数据时应先确认脱敏边界和操作权限,避免误改生产环境。

来源仓库内路径:skills/message

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/henkisdabro/wookstar-claude-plugins --skill me…;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

    它需要操作浏览器,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

message 怎么安装?

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

message 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。

message 需要什么权限?

权限级别是操作浏览器。该 Skill 可能涉及浏览器控制能力,使用时可能读取或操作网页内容,需要在受控环境中确认权限边界。

message 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。

message 用的人多吗?

公开数据是累计安装 436 次,近一周安装 18 次,下载 143 次,GitHub 51 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。

message 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

  • 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

  • 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
message

message 效率 · 本页

43651操作浏览器未明确有提醒

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

Message Drafts

Create rich text messages that paste perfectly into Gmail, Outlook, or convert to WhatsApp-formatted text. Fragments are written in Markdown - the build script converts to platform-specific HTML automatically.

Architecture

Claude writes              Build script assembles          Output (separate file)
name.fragment.md  --->     Markdown -> HTML                name.html
(10-30 lines)              Gmail transform (tags)          (self-contained preview
                           Outlook transform (styles)       with three body versions)
                           Inject into shell.html

The fragment is the source of truth. The assembled HTML is a derived output. Never edit the .html output directly.

Create Workflow

  1. Draft email content in conversation
  2. Write a .fragment.md file to data/writing/email_drafts/
  3. Run the assembler with --serve to produce the preview HTML and launch the server
uv run .claude/skills/message/scripts/assemble.py /path/to/name.fragment.md --serve

Run with run_in_background: true. The server picks a random free port and prints the URL. The preview auto-refreshes when the HTML file changes - no manual browser refresh needed.

Edit Workflow

When editing an existing email (argument provided, or user asks to change something):

  1. Read the small .fragment.md file
  2. Use the Edit tool to make targeted changes
  3. Re-run the assembler without --serve to rebuild the preview
uv run .claude/skills/message/scripts/assemble.py /path/to/name.fragment.md

The preview server is still running in the background from the create step. The assembler overwrites the HTML file atomically, and the browser auto-refreshes within 2 seconds. Do not re-launch the server on edits - the user keeps the same URL.

Fragment Format

Fragments are Markdown files with YAML frontmatter:

---
to: recipient@example.com
cc: optional@example.com
subject: Email Subject
---

Hi **Stuart**,

Here's the ~~old approach~~ new approach.

## Key Points

- First item
- Second item

> Quoted text from previous email

| Feature | Status |
|---------|--------|
| Auth    | Done   |

Inline `code` and [a link](https://example.com).

Cheers,
Henrik

Frontmatter Fields

FieldRequiredNotes
toYesComma-separated recipients
subjectYesEmail subject line
ccNoHidden in preview if empty
bccNoHidden in preview if empty

Formatting Reference

Standard Markdown maps to platform-specific HTML automatically:

What you wantWrite in Markdown
Bold**bold**
Italic*italic*
Strikethrough~~strikethrough~~
Heading## Heading
Bullet list- item
Numbered list1. item
Blockquote> quoted text
Inline code` code `
Code blocktriple backticks
Link[text](url)
Table`

For features not in standard Markdown, embed HTML directly (it passes through unchanged):

What you wantEmbed as HTML
Custom colour<span style="color: #c0392b;">red text</span>
Large text<font size="4">larger text</font>
Underline<u>underlined</u>
Indent<blockquote style="margin: 0 0 0 40px; border: none; padding: 0px;">indented</blockquote>

Example: User Asks for Colour and Strikethrough

User: "Write an email to Stuart. Strikethrough '$5000' and make his name red."

Claude writes .fragment.md:

---
to: akumanostuart@gmail.com
subject: Pricing Update
---

Hi <span style="color: #c0392b;">Stuart</span>,

The old pricing was ~~$5000~~ but the new rate is $3000.

Cheers,
Henrik

File Naming

data/writing/email_drafts/YYYY-MM-DD_recipient_subject.fragment.md   <- source
data/writing/email_drafts/YYYY-MM-DD_recipient_subject.html          <- assembled output

Examples:

  • 2026-02-12_veronika_audit-proposal.fragment.md
  • 2026-02-12_danielle_project-update.fragment.md

Platform Transforms

The build script produces three HTML versions from a single Markdown source:

Gmail (tag transform): Converts semantic HTML to Gmail-native elements. <p> becomes <div>, <strong> becomes <b>, <em> becomes <i>, headings become <font size> with <b>. Container gets Arial 13px. No custom colours added - only platform defaults.

Outlook (style injection): Adds inline styles to every element for Word's rendering engine. Aptos/Calibri 11pt, explicit color: #000000 on every text element, mso-line-height-rule for spacing control. No custom colours added - only platform defaults.

WhatsApp (text conversion): Strips HTML and converts to WhatsApp markdown - *bold*, _italic_, ~strikethrough~. Tables become pipe-separated text.

User-specified inline styles (colours, font sizes) are preserved through all transforms.

Preview

The assembled HTML has a Gmail/Outlook/WhatsApp mode toggle:

  • Gmail mode: Gmail-native HTML with Arial 13px container, Gmail action buttons (Compose in Gmail, Open Gmail Inbox, Copy for Gmail)
  • Outlook mode: Outlook-native HTML with full inline styles, Outlook action button (Copy for Outlook)
  • WhatsApp mode: Converts to WhatsApp-compatible plain text. Action buttons: Copy for WhatsApp, Send via WhatsApp

Instruct user:

  1. Click the URL printed by the server
  2. Select Gmail, Outlook, or WhatsApp mode
  3. Review the email preview
  4. Use the action buttons for their chosen client

Backward Compatibility

Legacy .fragment.html files still work. The assembler detects the extension and uses the appropriate parser. For HTML fragments, the same body is used for all three platform views (matching previous behaviour).

Structure

Follow this order in every email:

  1. Lead with the point - the request, update, instruction, or information the recipient needs to act on. This is why they opened the email.
  2. Supporting detail - context, logistics, or reference material that supports the lead.
  3. Warm close - any personal, relational, or appreciative content goes at the end, just before the sign-off. Gratitude, compliments, looking-forward sentiments, and rapport-building belong here - never at the top.

The recipient should know what the email is about within the first two lines. Warmth is the exit feeling, not the entrance.

Tone

  • Write naturally, as a human would compose
  • Avoid repetitive language from previous emails in the thread
  • Keep prose flowing, not overly structured
  • Match the recipient's formality level
  • Use British English spelling (colour, analyse, organise, behaviour, centre)

Henrik's Default Details

Use these when composing emails on Henrik's behalf:

PurposeEmail
Primary / From addressadmin@henriksoderlund.com
Google platform access requests (GA4, GTM, Google Ads, Search Console, etc.)admin@henriksoderlund.com

Sign-off block:

Henrik Soederlund
Independent Digital Consultant
admin@henriksoderlund.com

References

  • references/outlook-formatting.md - Outlook element styles and colour palette reference
  • references/formatting-rules.md - Gmail native HTML element reference

After Preview

Once the user has the preview URL, ask whether they'd like to run the humaniser skill on the message to remove any AI writing patterns before sending. Example prompt: "Would you like me to run the humaniser skill on this draft to make it sound more natural?"

Post-Draft Actions

After saving a draft, prompt Henrik: "Log this to the client comms log?" If yes, write the entry to data/consulting/clients/[slug]/comms/_comms-log.json via inline Python or the Edit tool (manage_comms.py is read-only - no --add flag).

平台分布

Codex

37.73%
按下载量换算54

Claude

28.31%
按下载量换算40

Cursor

19.28%
按下载量换算28

Gemini CLI

8.89%
按下载量换算13

安全审计

Gen Agent Trust Hub

通过

Socket

可疑

Snyk

通过

权限和风险

操作浏览器

该 Skill 可能涉及浏览器控制能力,使用时可能读取或操作网页内容,需要在受控环境中确认权限边界。

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