Slack Feedback

This Slack Feedback MCP Server collects product feedback from Bryan and teammates in the StartupOS Slack workspace. It exposes tools to pull stakeholder messages, fetch full conversation threads, and search feedback by keywords, making it easy to surface and analyze user input inside your Claude Code workflows.

How to use

You interact with this MCP server through an MCP client or Claude Code by calling its available tools. The server exposes three core tools you can use in natural-language prompts: get_stakeholder_feedback to pull messages with date filtering, get_thread_context to retrieve full discussion threads, and search_feedback to locate messages by keywords. When you issue a query, keep in mind the following patterns to accomplish common tasks quickly: pull recent feedback from Bryan in a specific channel, fetch the full thread for a given message, or search for feedback containing a keyword like a feature name or an issue.

How to install

Prerequisites you need before starting:

• Node.js installed on your machine (Node.js 14+ is typical for MCP projects).

• npm available on your system.

Configuration and operation

Set up the Slack integration and the MCP endpoint as shown below to enable Claude Code to reach your MCP server.

{
  "mcpServers": {
    "slack_feedback": {
      "type": "http",
      "url": "https://YOUR-RAILWAY-URL.up.railway.app/sse",
      "args": []
    }
  }
}

Testing the setup

After deploying, restart Claude Code and try these prompts to verify everything is wired correctly: pull feedback from Bryan from the last 48 hours, search for feedback mentioning a keyword like authentication, or get the full thread for a specific message using its thread timestamp.

Security and access

Keep your tokens secure. Do not commit credentials to version control. The Slack bot token should have only read permissions for the channels and messages you need.

Troubleshooting

If you encounter a "Channel not found" error, ensure the bot is invited to the channel and the Channel ID is correct. For a "Not authorized" error, verify that all required Slack scopes are granted and reinstall the app after updating permissions. If no messages are returned, confirm Bryan’s user ID is correct and that there are messages in the selected time range.

Notes

The MCP is designed to be extended. If you need to enable summarization or categorization later, you can add environment variables and adjust the code to activate those features while keeping raw messages available for flexibility.

What you get from this server

Three MCP tools are available: get_stakeholder_feedback, get_thread_context, and search_feedback. They enable you to collect, contextualize, and explore feedback from the StartupOS Slack channel with flexible filtering and retrieval options.

Available tools

get_stakeholder_feedback

Pull messages from the feedback channel with optional time range and stakeholder filters to surface the most relevant feedback.

get_thread_context

Retrieve the full thread for a given message, including all replies within the channel.

search_feedback

Search messages by keywords across the feedback channel to locate specific feedback quickly.