Skool MCP
Gives Claude access to your Skool community, enabling reading and writing posts, comments, members, classroom content, and DMs using your own login.
README
Skool MCP
Give Claude access to your Skool community. Posts, comments, members, classroom, and your DM inbox.
Ask Claude things like "what did people post this week?", "draft a reply to the last 5 DMs", or "pull my member list into a spreadsheet". It runs on your computer, using your own Skool login.
Unofficial. Not built by Skool, not affiliated with Skool.
What you can do with it
13 tools. The read ones are safe to fire off. The write ones post publicly under your name, so Claude should confirm the wording with you first.
| Tool | What it does | Read or write |
|---|---|---|
skool_whoami |
Confirms your Skool login works and lists the communities you belong to. Run this first when something breaks. | read |
skool_list_posts |
Lists recent posts in your community feed: title, snippet, comment and upvote counts, author. | read |
skool_get_post |
Reads one full post and its comments. | read |
skool_create_post |
Publishes a new post to your community feed. | write |
skool_comment_on_post |
Adds a comment to a post. | write |
skool_list_members |
Lists members of your community: id, name, email, short bio. | read |
skool_list_courses |
Lists the classroom courses in your community. | read |
skool_get_lesson |
Reads a course's modules and lessons, and the content of one lesson. | read |
skool_list_conversations |
Your DM inbox: unread counts and last-message previews. | read |
skool_get_conversation |
Reads the messages in one DM conversation. | read |
skool_send_message |
Sends a DM into an existing conversation. | write |
skool_mark_read |
Marks a DM conversation as read. | write |
skool_start_chat |
Opens a new DM with a member. | write |
Before you start
- macOS or Windows. (Linux is untested.)
- Claude Code or Claude Desktop.
- A Skool account that is actually a member of the community you want to connect.
- That's it. You do not need Python installed. The setup installs everything it needs.
Install
Step 1. Get the folder onto your computer. Either way works.
Option A, if you have git:
cd ~
git clone https://github.com/Solnest-AI/skool-mcp-community.git skool-mcp
That drops it at ~/skool-mcp, which is exactly where you want it.
Option B, the ZIP:
- Download the ZIP (on the repo page: green Code button, then Download ZIP).
- Unzip it.
- Rename the folder to
skool-mcpand move it so it sits directly in your home folder.
Step 2. Put it in the right place. This matters more than it sounds like it should.
- Mac:
~/skool-mcp(that is/Users/yourname/skool-mcp) - Windows:
C:\Users\yourname\skool-mcp
Not Documents. Not Desktop. Not Downloads. macOS locks down those folders, and background tasks that try to read them fail in a way that is genuinely annoying to debug. Home folder, done.
Step 3. Point Claude at the folder and say "set this up".
In Claude Code:
cd ~/skool-mcp
claude
Then type: Set this up.
Claude reads SETUP.md and runs the whole thing: installs the tools, opens a browser so you can
log into Skool yourself, verifies the connection, registers the server, and tests it. It stops and
asks you when it needs you.
Setting up for Claude Desktop instead? Same thing. Run the setup from Claude Code (or by hand), and at the registration step tell Claude you want to use it in Claude Desktop.
How this works and what to know
Read this part. It is short and it is the honest version.
Skool has no official API. There is no developer program, no API key, no supported way in. So this does what your browser does: it uses your own logged-in session to talk to the same internal endpoints the Skool website uses.
Your login stays on your machine. During setup, a real browser window opens and you log into
Skool yourself, exactly like normal (Google sign-in and 2FA both work). The script reads the
session cookie out of that browser and writes it to a local file called .env. Nothing is sent to
Solnest, to Anthropic, or anywhere else.
Never paste your cookie into a chat. Not to Claude, not to me, not to anyone in the community being helpful. That cookie is your entire Skool account: your posts, your DMs, your billing, your members. Anyone holding it is logged in as you. Setup is built so the cookie never has to appear in a conversation, and Claude is instructed to refuse if you offer it.
.env is your password file. Treat it exactly like one. Don't open it in a screenshare, don't
screenshot it, don't paste it anywhere, don't drop it in Drive or Slack. The bundle is already set
up to keep it out of any code you upload, so you mostly just have to not go out of your way to
share it. If you ever think it leaked, log out of Skool everywhere and re-run setup.
The tools act as you. A post created by skool_create_post is your post. A DM is from you. It
looks identical to typing it yourself, because functionally it is. Have Claude read the draft back
to you before anything goes out.
Only use it on communities you actually belong to. These are your own credentials against your own memberships. Don't get cute.
Skool can break it. No official API means no stability promise. If Skool changes their site, tools can stop working overnight. When something breaks, run the doctor first:
cd ~/skool-mcp
uv run doctor.py
It tells you exactly which piece is broken and how to fix it. It never prints your cookie.
Your session expires eventually. The Skool login token lasts about a year. The AWS firewall
token bundled with it lasts about 4 days, and that is the one that quietly kills your tools. Setup
offers an optional background job that refreshes it for you. If you skip it, just re-run
uv run connect_skool.py whenever the tools start failing.
Getting help
Run uv run doctor.py, then paste the output into Claude and ask it to fix what failed. The output
is safe to share: it prints check results, cookie lengths, and expiry dates, never the cookie itself.
Still stuck? Post in the community with the doctor output. Do not post your .env.
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。