discord-mcp
Enables Claude to read Discord message threads using a personal user token, so it can fetch thread content from any server you're a member of.
README
discord-mcp
A personal MCP (Model Context Protocol) server that lets Claude read Discord message threads. Give Claude a Discord message URL and it can fetch the thread content directly.
What it does
Exposes a single read_discord_thread tool to Claude that:
- Fetches the anchor message and all subsequent messages in a channel thread
- Supports pagination for long threads via
after_idandbatch_size - Skips system messages and embeds (empty-content messages)
- Returns structured data: author, timestamp, content, and pagination state
This uses your personal Discord account token, so it can read any server you're a member of — no bot invite required.
Obtaining your Discord auth token
Warning: Your user token is equivalent to your Discord password. Never share it or commit it to version control.
- Open Discord in your browser (discord.com)
- Open DevTools (
F12orCtrl+Shift+I) - Go to the Network tab
- In Discord, navigate to any channel or send any message to trigger a network request
- Filter requests by
XHRor search formessages - Click any request to
discord.com/api/v9/ordiscord.com/api/v10/ - In the Request Headers section, find the
Authorizationheader — its value is your token
Alternatively, in the Console tab you can run:
webpackChunkdiscord_app.push([[Math.random()],{},r=>{m=[];for(let c in r.c)m.push(r.c[c])}]);
m.find(m=>m?.exports?.default?.getToken).exports.default.getToken()
Copy the token value (no Bot prefix — that's only for bot accounts).
Setup
Prerequisites: Python 3.10+, uv
cd /path/to/discord-mcp
uv venv
uv pip install -r requirements.txt
Create a .env file (see .env.example):
DISCORD_USER_TOKEN=your_token_here
Testing
Run the unit tests:
.venv/bin/pytest
Run the smoke test with a real Discord URL:
.venv/bin/python smoke_test.py
Registering with Claude Code
Register the server globally so it's available in all your Claude Code sessions:
claude mcp add --scope user discord \
/path/to/discord-mcp/.venv/bin/python \
/path/to/discord-mcp/server.py
Verify it's connected:
claude mcp get discord
You should see Status ✔ Connected.
Registering with Codex
Register the server with Codex as a local stdio MCP server:
codex mcp add discord -- \
/path/to/discord-mcp/.venv/bin/python \
/path/to/discord-mcp/server.py
Verify that it is configured:
codex mcp get discord
codex mcp list
Restart Codex or open a new session after adding the server. In the Codex
terminal UI, run /mcp to confirm that discord and its
read_discord_thread tool are available.
Usage
Once registered, ask Claude or Codex things like:
"Read this Discord thread: https://discord.com/channels/123/456/789"
"Summarize the conversation starting at this message: https://discord.com/channels/..."
The assistant will call read_discord_thread and have access to the message content. For long threads, it can paginate by passing the returned last_id as after_id in subsequent calls.
How it works
- Uses Discord's REST API v10 with your user token
- Fetches the anchor message via
?around=<id>&limit=1(the direct endpoint is bot-only) - Paginates forward with
?after=<id>&limit=<batch_size> - Runs as a stdio MCP server — Claude Code or Codex spawns it as a subprocess when needed
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。