TL;DR:
Connecting Claude Desktop or any AI assistant to Discord through a manual, self-hosted MCP server means managing Docker containers, Python environments, and bot tokens that break the moment your machine sleeps.
Composio eliminates that overhead with a managed MCP server that connects your AI to Discord with secure In-Chat Auth, sandboxed execution, and access to 1,000+ pre-built toolkits.
Our free tier gives you 20,000 tool calls per month with no credit card required.
Most guides for connecting an AI assistant to Discord assume you're comfortable maintaining a local Python environment all day. This guide covers three paths: ChatGPT Work and Claude Cowork both connect through a UI with no configuration files involved. Cursor requires editing a JSON config and is aimed at developers already working inside the editor. All three use Composio's managed MCP server, so there's no custom integration code to write and no self-hosted server to maintain.
The practical result: your AI can draft announcements, pull channel history, manage roles, and trigger actions in other apps, all from a single connection that stays live without you babysitting it.
How to integrate Discord MCP with ChatGPT Work
Prerequisites
A ChatGPT account with access to ChatGPT Work
A Composio account
Access to the Discord server you want to connect
Permission to approve the requested Discord access
Step 1: Open the MCP settings
Open Settings in ChatGPT Work and select Plugins. Go to the MCP tab.
Step 2: Add the Composio server
Click Add server, enter Composio as the name, and select Streamable HTTP as the type. Add https://connect.composio.dev/mcp as the server URL, then click Save.
Step 3: Authenticate your Composio account
Click Authenticate beside the newly added Composio server. Sign in to Composio and select Allow access to connect it to ChatGPT Work.
Step 4: Connect and use Discord
Ask ChatGPT Work to perform a task involving Discord, then sign in to Discord and approve the requested permissions when prompted. You can then use natural-language requests to list your servers, view connected accounts, or retrieve your Discord profile information.
How to integrate Discord MCP with Claude Cowork
Prerequisites
The Claude Desktop app with access to Cowork
A Composio account
A Discord account
Permission to approve the requested Discord access
Step 1: Open the connector settings
Open Claude Desktop and select Customize from the left sidebar. Choose Connectors, then click the + icon.
Step 2: Add the Composio MCP server
Select Add custom connector and enter https://connect.composio.dev/mcp as the server URL. Click Connect to continue.
Step 3: Authorize Composio
Claude will open a browser window where you can authorize Composio. Review the request and approve access to complete the connection.
Step 4: Connect Discord
Return to Cowork and ask it to perform a Discord task, such as listing the servers you belong to. Sign in to Discord and approve the requested permissions when prompted.
How to integrate Discord MCP with Cursor
This section is aimed at developers. It requires editing a JSON configuration file in your project root or home directory.
Prerequisites
The Cursor desktop application
A Composio account
A Discord account
Permission to authorize access to Discord
Step 1: Install Composio in Cursor
Click Install in Cursor on the Composio page to add the MCP server automatically. This is the quickest installation option.
Step 2: Alternatively, add Composio manually
Open .cursor/mcp.json in your project root or ~/.cursor/mcp.json for a global setup. Add Composio as an HTTP server using https://connect.composio.dev/mcp.
Step 3: Authorize Composio
Restart Cursor and open the MCP Tools settings. Click Connect beside Composio and complete the authorization in the browser window.
Step 4: Connect your Discord account
Ask the Cursor agent to connect to Discord or complete a Discord-related task. Follow the prompt to authenticate your Discord account and authorize access.
Ways Discord MCP can help with everyday tasks
Catch up on busy channels
Active channels generate a lot of noise. When hours or days of messages have piled up, spotting the useful updates means reading through everything—general chatter included. Discord MCP reads through recent activity and pulls out the points that matter. You get a clear picture of what changed, what was decided, what's still blocked, and whether anyone is waiting on a response.
Example prompt:“Summarize everything posted in the #team-updates channel since Monday and give me a clear breakdown of the main decisions, deadlines, blockers, assigned tasks, and anything that still needs a response.”
Find information from older conversations
Decisions made in Discord don't stay easy to find. Once the conversation moves on and new messages pile up, surfacing a specific topic means manually combing through channels, threads, and old replies. Discord MCP can locate the relevant messages and explain what was discussed, what conclusion the team reached, and whether any questions were left open.
Example prompt:“Search our Discord server for previous conversations about the pricing update and summarize what was agreed, what concerns were raised, who was involved, and what still needs to be decided.”
Turn discussions into a clear task list
Work gets agreed on in Discord all the time without ever being formally recorded anywhere else. Tasks get missed or misunderstood because there's no structured record of what was said. Discord MCP can review the conversation and turn it into a practical list of next steps—owner, deadline, status, and a note for anything that still needs clarification before the work can move forward.
Example prompt:“Review the latest messages in #project-alpha and turn the discussion into a task list with an owner, deadline, priority, status, and a note for anything that is still unclear.”
Review customer and community feedback
Feedback channels mix feature requests, bug reports, questions, complaints, and praise into the same stream of messages. The useful signals are in there, but they're hard to see when every message sits on its own. Discord MCP can review a larger group of messages, combine similar comments, and surface the main themes. Your team gets a clear view of what people ask for most often and where the biggest problems are concentrated.
Example prompt:“Review all messages posted in #customer-feedback this week, group similar comments together, and summarize the most common bugs, feature requests, complaints, questions, and positive feedback.”
Draft replies using the full conversation
Writing a useful reply often requires more than reading the latest message, especially when several people have already responded or the issue has developed over a longer thread. Discord MCP can review the full context and identify what has already been explained. From there, it helps you draft a response that answers the question clearly, avoids repeating earlier points, and gives the person a sensible next step.
Example prompt:“Read the full conversation in the latest #support thread and draft a clear reply that answers the user’s question, acknowledges what has already been tried, and explains what they should do next.”
Create weekly or project updates
Putting together a weekly update can take longer than expected when information is spread across several channels and different team members share progress in different formats. Discord MCP can collect those updates and organize them into a consistent summary. The result is a report covering completed work, current priorities, blockers, upcoming deadlines, and any decisions that need wider team attention.
Example prompt:“Create a weekly project update using the messages from #launch-team and include what was completed, what is currently in progress, any blockers, upcoming deadlines, important decisions, and the next steps for the team.”
Spot urgent issues before they get buried
Support, bug, and incident channels can move quickly, which means a serious customer problem may be pushed out of view by newer messages before the right person notices it. Discord MCP can review recent reports and group messages describing the same problem. It flags issues that are still unresolved and helps you understand which ones need immediate attention based on impact, frequency, and how long they've been open.
Example prompt:“Review the latest messages in #incident-reports, group duplicate issues together, and rank the unresolved problems by urgency based on customer impact, number of reports, and how long each issue has been open.”
Composio acts as the integration layer between Discord and tools like ChatGPT, Claude, and Cursor, handling the connection, authentication, and tool access needed to make these workflows possible. Instead of building and maintaining the integration yourself, you can use Composio to connect Discord once and start working with messages, channels, threads, and server data from the AI tool you already use.
Create a free Composio account to get started with 20,000 tool calls per month, no credit card required.
FAQs
Can I use Discord MCP with custom GPTs or LangChain agents?
Not directly for custom GPTs. Custom GPTs built in the GPT Builder use OpenAPI-based Actions rather than MCP, so they can't connect to the Composio MCP server directly. Use ChatGPT Work instead, which supports MCP natively as shown in the steps above. LangChain and other frameworks are fully supported, though. We support all major AI frameworks through provider packages: LangChain (Python and TypeScript), CrewAI, LlamaIndex, AutoGen, and Mastra, plus any MCP-compatible client. Each MCP-compatible framework connects through your own Composio MCP endpoint, so Discord tools are available regardless of the framework you're running.
How does Composio protect my Discord server data?
Composio applies zero-day log retention by default, meaning tool call data is not stored after execution. Tool call execution happens inside a sandboxed environment on Composio's infrastructure, isolated from your local machine. Audit logs record the action taken and when it happened, but response payload content is not retained beyond execution.
How do I remove Discord MCP access?
You have two options. To stop agent-initiated tool calls, disconnect Discord from the Composio dashboard under Connected Accounts—this immediately revokes Composio's access to your server without removing anything from Discord itself. To revoke access at the Discord level, go to User Settings → Authorized Apps in Discord and remove Composio from the list of approved applications.
What's the difference between MCP and a standard Discord webhook?
Webhooks are one-way: they either push messages into Discord or push event notifications from Discord to an external URL, but they can't pull data, make decisions, or reason about what they retrieve. MCP is a bidirectional protocol: your AI can read channel history, evaluate content, and post a response in the same workflow.
Key terms
MCP server: A service that exposes tools, resources, and prompts to AI assistants through the Model Context Protocol, enabling bidirectional communication between the assistant and external applications.
OAuth2 scope: A permission definition in Discord's authorization system that determines which actions and data your bot can access, such as reading message history or managing roles.
Sandboxed execution: An isolated execution environment where Composio runs tool calls on its own infrastructure rather than your local machine, preventing code execution risks from affecting your local environment.
Tool Router: A Composio feature that dynamically routes agent requests to the appropriate connected service based on user authentication, eliminating the need for conditional logic in agent code.
In-Chat Auth: A Composio authentication system that allows agents to prompt users for credentials inline during a conversation when a tool requires authentication, without interrupting the workflow or requiring manual configuration file edits.