claude-code-discord

Advantage	Details	Status
Use Claude Code Anywhere	Host locally (VM / Docker / cloud) and send commands via the Discord API	✅
Centralized collaboration	Run commands and discuss results where your team already communicates	✅
Branch-aware organization	Maps Git branches to channels/categories so feature work stays separated	✅
Immediate, shareable feedback	Execute /git, /shell, or /claude and get outputs directly in-channel	✅
Reduced context switching	Keep actions, logs, and discussion together — less switching between terminal and chat	✅
Role-based access control	Restrict destructive commands (/shell, /git, worktree ops) to specific Discord roles	✅
Non-developer friendly	PMs, QA, and stakeholders can trigger checks and view results without repo access	✅
Automatable touchpoint	Integrate with CI/webhooks to triage issues and run fixes from chat	✅
Local hosting & security	Keep keys and code on your infra while exposing a controlled interface through Discord	✅
Audit trail & accountability	Channel history provides an easy-to-search record of who ran what and when	✅

Start Here If You Have These: DISCORD_TOKEN APPLICATION_ID

• Quick Start
• Command List 48 Commands

Tutorial If you dont know or have them follow these first then come back:

• How To Setup Discord Bot?

---

Preview:

Quick Start

• Install Deno Can Be Done Via Denos Website Or Commands Under:

# Linux/MacOS
curl -fsSL https://deno.land/install.sh | sh

# Windows|Powershell
irm https://deno.land/install.ps1 | iex

Clone the project:

git clone https://github.com/zebbern/claude-code-discord.git
cd claude-code-discord

Install claude If you dont have it and login:

npm install -g @anthropic-ai/claude-code
claude /login

Required environment variables

# Linux
export DISCORD_TOKEN="your-discord-bot-token"
export APPLICATION_ID="your-discord-app-id"

# Windows Terminal 
set DISCORD_TOKEN=your-discord-bot-token
set APPLICATION_ID=your-discord-app-id

# Windows Powershell
$env:DISCORD_TOKEN = "your-discord-bot-token"
$env:APPLICATION_ID = "your-discord-app-id"

Now Run the discord Bot

If you get not a git directory just run git init or git add .

# Option 1: Just run the bot
deno run --allow-all index.ts

# Option 2: Run the bot with extra terminal output (For Dev)
deno run --allow-all index.ts --watch

# Option 3: Run the bot and get pings when claude is done
./index.ts --category myproject --user-id Your_Discord_User_ID_Here

# Can also be ran like this: 
deno run --allow-all index.ts --category yourproject --user-id Your_Discord_User_ID_Here

You can run without --user-id Your_Discord_User_ID_Here if you dont want to be notified when claude finishes

Setup Discord Bot

1. Create a Discord Application

Note

• Go to the Discord Developer Portal
• Click New Application
• Give your application a name (e.g., ClaudeCode)
• Click Create

2. Copy Application ID (Needed For Config)

Note

• Go to the General Information → Copy Application ID section

3. Create a Bot User

Note

• In your application, go to the Bot section in the left sidebar
• Click Add Bot
• Under Token click Copy to copy your bot token (keep this secure!)
• Click Save Changes

4. Invite the Bot to Your Server

Note

• Go to the OAuth2 → URL Generator section
• Under Scopes select:

+ | bot
+ | applications.commands

• Under Bot Permissions select:

+ | Send Messages
+ | Use Slash Commands
+ | Read Message History
+ | Embed Links

Copy the generated URL and open it in your browser Select your Discord server and authorize the bot

Command List

(48 Commands)

Core Claude (3)

• /claude, /continue, /claude-cancel

Enhanced Claude (4)

• /claude-enhanced, /claude-models, /claude-sessions, /claude-context

Development Tools (7)

• /claude-explain, /claude-debug, /claude-optimize, /claude-review
• /claude-generate, /claude-refactor, /claude-learn

New Features (3)

• /todos - Task management with API rate limits
• /mcp - Model Context Protocol servers
• /agent - 7 specialized AI agents

Settings (4)

• /settings - Unified settings (NEW)
• /claude-settings, /output-settings, /quick-model

Git Operations (6)

• /git, /worktree, /worktree-list, /worktree-remove
• /worktree-bots, /worktree-kill

Shell Management (4)

• /shell, /shell-input, /shell-list, /shell-kill

System Monitoring (10)

• /system-info, /processes, /system-resources, /network-info
• /disk-usage, /env-vars, /system-logs, /port-scan
• /service-status, /uptime

Utilities (4)

• /status, /pwd, /shutdown, /help

Agent System (3)

• /agent with 7 specialized agents:
  • Code Reviewer, Software Architect, Debug Specialist
  • Security Analyst, Performance Engineer, DevOps Engineer, General Assistant

Thinking Mode Options ✨

• none - Standard Claude responses
• think - Step-by-step reasoning mode
• think-hard - Deep analysis and reasoning
• ultrathink - Maximum depth thinking for complex problems

Operation Mode Options ✨

• normal - Standard operation with user confirmation
• plan - Planning mode without execution
• auto-accept - Automatically apply suggested changes
• danger - Unrestricted mode (high risk)

/todos Command ✨

• Action types: list, add, complete, generate, prioritize, rate-status
• Priority levels: low, medium, high, critical
• Rate limit awareness - Supports Anthropic API tiers including exceeds_200k_tokens
• Token estimation - Calculates estimated token usage
• Auto-generation - Generate todos from code files

/mcp Command ✨

• MCP server management - Model Context Protocol integration
• Server types: local, http, websocket, ssh
• Actions: list, add, remove, test, status
• Connection testing - Verify MCP server connectivity

/agent Command ✨

• Specialized AI agents for different development tasks:
  • Code Reviewer - Quality analysis and security
  • Software Architect - System design and architecture
  • Debug Specialist - Bug analysis and troubleshooting
  • Security Analyst - Vulnerability assessment
  • Performance Engineer - Optimization and profiling
  • DevOps Engineer - Deployment and infrastructure
  • General Assistant - Multi-purpose development help
• Risk levels - Low/Medium/High risk classification
• Session management - Persistent agent conversations
• Context awareness - Include system info and files