Connect Commslayer to Claude or ChatGPT. Manage tickets, update AI guidance, and write help articles through conversation.
What is MCP?
MCP (Model Context Protocol) lets AI assistants connect to external tools. Commslayer's MCP server gives Claude and ChatGPT direct access to your helpdesk.
Before you start
Early beta. Expect rough edges.
Paid plans required: Claude Pro or Team. ChatGPT Plus, Pro, or Team.
Admin access required: You must be an administrator on your Commslayer account. Managers and agents cannot connect.
Use Claude. ChatGPT technically works but is almost unusable for complex tasks. Claude handles helpdesk work far better.
Connect Claude (Web)
-
Open claude.ai and go to Settings → Connectors
-
Click Add custom connector
-
Enter:
- Name:
Commslayer - URL:
https://app.commslayer.com/api/v1/mcp
- Name:
-
Click Add
-
Sign in to Commslayer when prompted
Click Configure to set tool permissions (auto-approve, ask each time, or never allow).
Connect Claude Code
Option 1: Ask Claude to set it up
Paste this into Claude Code:
Configure Commslayer MCP with URL: https://app.commslayer.com/api/v1/mcp
Complete authentication when prompted.
Option 2: Manual setup
-
Run
/mcpto open MCP configuration -
Add a server with URL:
https://app.commslayer.com/api/v1/mcp -
Complete authentication when prompted
Verify: Ask "What inboxes do I have in Commslayer?"
Connect ChatGPT
Requirements: Plus, Pro, or Team account. Developer mode enabled.
Enable developer mode: Settings → Apps → toggle on Developer mode
Create the app:
-
Click your profile avatar → Settings
-
Go to Apps
-
Click Advanced settings → Create app
-
Enter:
- Name:
Commslayer - MCP Server URL:
https://app.commslayer.com/api/v1/mcp - Authentication:
OAuth
- Name:
-
Check "I understand and want to continue" → Create
Connect to Commslayer:
-
Click Connect on the app details page
-
Review permissions → Continue to Commslayer
-
Sign in to Commslayer → Authorize
ChatGPT asks for permission the first time you use Commslayer tools in a conversation.
What you can do
Conversations
- Search and filter by status, date, labels, or channel
- View full history with AI agent activity
- Resolve, snooze, or reassign
- Add or remove labels
Help center
- Create articles from customer questions
- Turn support conversations into documentation
- Publish directly
AI agent configuration
- View and update guidance rules
- Create guidance for specific scenarios
- Manage canned responses
Testing
- Test AI responses to specific conversations
- Batch test guidance changes
- Validate before enabling
Example prompts
Find automation opportunities:
Look through our last 200 conversations. Find patterns where
customers ask similar questions that the AI agent could handle
but doesn't. Suggest new guidance rules.
Write help articles from conversations:
Review recent conversations and find topics we should document.
Ask me questions about anything you need to know, then create
the articles.
Audit AI knowledge for conflicts:
Review our AI agent's guidance rules, canned responses, and
help articles. Find conflicts or inconsistencies where
different sources give different answers.
Find ad hooks from comments:
Look through our Meta inbox comments from the last 30 days.
Find patterns: what objections come up? What questions do
people ask? What praise do they give? Use their exact words
to suggest new ad hooks.
Limits
- Daily messages: 2,000 per account
- Results per page: 50 max
- Total results: 500 max per query
- Batch requests: 20 max
Security
- OAuth authentication—no tokens to manage
- Access scoped to your account only
- Removing someone from your account revokes access immediately
- Only administrators can connect—managers and agents cannot use MCP