Slack Explorer MCP

io.github.shibayu36由 GitHub 账号 shibayu36 发布v0.12.0更新于 Oct 24, 2026

MCP server specialized in retrieving information from Slack messages and threads

已验证STDIO仅桌面知识与记忆沟通与协作

概览

AI 生成的概览

只读的 Slack 检索服务器,让助手搜索该用户令牌可见的消息、线程、文件、画布和用户资料。

可以做什么
  • 按频道、发送者、日期范围、表情回应、附件和排序方式搜索消息
  • 分页读取某个消息线程的全部回复
  • 批量查询用户资料,并按显示名称查找用户
  • 搜索画布、PDF、图片、表格等文件,并获取画布的 HTML 内容
适用场景
适合让助手翻查 Slack 历史记录:在频道和私聊中找回过去的讨论、决定或附件,或完整拉取某个线程及相关人员信息,而无需离开聊天客户端。
运行要求
以 Docker 容器(ghcr.io/shibayu36/slack-explorer-mcp)在本地运行,通常使用 stdio。需要一个配置了频道、私聊、搜索、用户和文件历史权限的 Slack 应用,以及 Slack 用户 OAuth 令牌(xoxp-),通过 SLACK_USER_TOKEN 传入;设置 TRANSPORT=http 时改用 X-Slack-User-Token 请求头。
安装前请注意
该 Slack 用户令牌可读取该用户可见的一切内容,包括私有频道和私聊,应视为敏感密钥。HTTP 模式下令牌经请求头传输,且服务器默认绑定所有网络接口,需限制主机和端口。检索到的消息、文件和用户资料(含邮箱地址)会发送给模型提供方。

安装

在 SourceWeft 中

  1. 打开 控制台中的 Slack Explorer MCP,将其添加到工作区。
  2. 为需要使用其工具的对话启用该服务。

Desktop only,通过 STDIO。 STDIO 服务会启动本地进程,因此需要 SourceWeft 桌面宿主。

其他 MCP 客户端

参照 仓库 中的启动说明。

README

Slack Explorer MCP Server

A Model Context Protocol (MCP) server specialized in retrieving information from Slack messages and threads. It provides tools to access messages that the authenticated user can view using a User Token (xoxp).

Available Tools

  • Message Search (search_messages)

    • Search Slack messages with advanced filtering options. You can search by channel, user, date range, and specific features (reactions, files, etc.).
    • Parameters
      • query: Basic search query. Use dedicated fields for inclusion modifiers (bare modifiers in this field are rejected). Exclusion modifiers prefixed with - (e.g., -in:#channel) ARE allowed here
      • in_channel: Filter by channel name (e.g., "general", "team-dev")
      • from_user: Search messages from specific user (User ID)
      • with: Search DMs/threads with specific users (array of User IDs)
      • before, after, on: Date range filtering (YYYY-MM-DD format)
      • during: Period specification (e.g., "July", "2023")
      • has: Messages containing specific features (emoji, "pin", "file", "link", "reaction")
      • hasmy: Messages where you reacted with specific emoji
      • sort: Sort method ("score" or "timestamp")
      • count: Number of results per page (1-100, default: 20)
      • page: Page number (1-100, default: 1)
  • Thread Replies (get_thread_replies)

    • Get all replies in a message thread. Supports pagination for efficiently handling large numbers of replies.
    • Parameters
      • channel_id: Channel ID (required)
      • thread_ts: Parent message timestamp (required)
      • limit: Number of replies to retrieve (1-1000, default: 100)
      • cursor: Pagination cursor
  • User Profiles (get_user_profiles)

    • Get profile information for multiple users in bulk. Retrieve display names, real names, email addresses, and other profile information by specifying a list of user IDs.
    • Parameters
      • user_ids: Array of user IDs (required, max 100)
  • Search Users by Display Name (search_users_by_name)

    • Search users by their display name. Supports both exact match and partial match search with case sensitivity.
    • Parameters
      • display_name: Display name to search for (required)
      • exact: Enable exact match search
  • File Search (search_files)

    • Search for files such as canvases, PDFs, and images. You can filter by file type, channel, user, and date range.
    • Parameters
      • query: Basic search query. Use dedicated fields for inclusion modifiers (bare modifiers in this field are rejected). Exclusion modifiers prefixed with - (e.g., -type:pdfs) ARE allowed here
      • types: Filter by file types (e.g., ["canvases", "pdfs"]). Available types: lists, canvases, documents, emails, images, pdfs, presentations, snippets, spreadsheets, audio, videos
      • in_channel: Filter by channel name (e.g., "general", "team-dev")
      • from_user: Search files from specific user (User ID)
      • with_users: Search files in DMs/threads with specific users (array of User IDs)
      • before, after, on: Date range filtering (YYYY-MM-DD format)
      • count: Number of results per page (1-100, default: 20)
      • page: Page number (1-100, default: 1)
  • Canvas Content (get_canvas_content)

    • Get HTML content of Slack canvases. Retrieve canvas content by specifying canvas IDs.
    • Parameters
      • canvas_ids: Array of canvas IDs (required, max 20)

Setup

Getting a Slack User Token

  1. Create an app at Slack API
  2. Add the following User Token Scopes in OAuth & Permissions:
    • channels:history - For public channels
    • groups:history - For private channels
    • im:history - For direct messages
    • mpim:history - For group direct messages
    • search:read - For message search
    • users.profile:read - For user profiles
    • users:read - For user information
    • files:read - For file content access
  3. Install the app to your workspace
  4. Get the User OAuth Token (starts with xoxp-)
    • Tip: To use with multiple users in the same workspace, add them as Collaborators and have each user reinstall from OAuth & Permissions to get their own User OAuth Token

MCP Server Configuration

  1. Configure mcp.json

    json
    {  "mcpServers": {    "slack-explorer-mcp": {      "command": "docker",      "args": ["run", "-i", "--rm", "--pull", "always",        "-e", "SLACK_USER_TOKEN=xoxp-your-token-here",        "ghcr.io/shibayu36/slack-explorer-mcp:latest"      ]    }  }}

    If you're using Claude Code:

    bash
    claude mcp add slack-explorer-mcp -- docker run -i --rm --pull always \  -e SLACK_USER_TOKEN=xoxp-your-token-here \  ghcr.io/shibayu36/slack-explorer-mcp:latest
  2. Use the agent to perform Slack searches

    Examples:

    • "Search for meeting-related messages in the general channel from last week"
    • "Find messages from @john.doe about 'project'"
    • "Get all thread replies for this post"

Usage

Common Search Patterns

  • Search in a specific channel

    Search for "release" messages in the general channel
  • Search messages from a specific user

    Search for yesterday's messages from @john.doe
  • Search messages with reactions

    Search for messages with :fire: reactions
  • Search messages you reacted to

    Search for messages where you reacted with :eyes:
  • Search messages with file attachments

    Search for messages with file attachments

Using as Streamable HTTP Server

By default, the server uses stdio for MCP communication. You can start it as a Streamable HTTP server by setting the TRANSPORT=http environment variable. In HTTP mode, pass the Slack token using the X-Slack-User-Token header.

Starting the server:

bash
# Start HTTP server (default: all interfaces 0.0.0.0, port 8080)docker run -i --rm --pull always \  -e TRANSPORT=http \  -p 8080:8080 \  ghcr.io/shibayu36/slack-explorer-mcp:latest
# Start with custom host and portdocker run -i --rm --pull always \  -e TRANSPORT=http \  -e HTTP_HOST=127.0.0.1 \  -e HTTP_PORT=9090 \  -p 9090:9090 \  ghcr.io/shibayu36/slack-explorer-mcp:latest

来源:README.md,提交 ea0b44d

工具

0
SourceWeft 尚未在沙箱中启动过该软件包,因此其工具未知。

版本历史

1
  1. v0.12.0最新Sep 16, 2026

评分

已满足适用检查项的 50%

该评分衡量此收录条目的完整程度,依据注册中心条目、端点与公开来源计算得出。它不是对服务器本身的评价,也不表示服务器安全或当前可用:可连通性是带日期的检查结果,评分也不包含安装量或用户评价。

  • 近 30 天内可连通25 分

    请保持端点能够响应 MCP 握手;SourceWeft 每 30 天探测一次。

  • 已声明认证方式10 分

    请在注册中心条目中声明凭据请求头,或保持端点能够响应探测,以便判定其认证方式。

  • 已列出工具15 分

    请让服务器在无凭据时响应 tools/list,或发布可在沙箱中无凭据启动的软件包。

  • 描述必需15 分
  • 图标10 分
  • 文档15 分
  • 许可证10 分