Connecting Asset Management for Jira to AI assistants via MCP

Asset Management for Jira offers a Model Context Protocol (MCP) server that lets you connect your inventory to AI assistants like ChatGPT and Claude.

Once connected, you can search your inventory, check assets and accessories in and out, manage licences, and link assets to Jira issues, all through natural conversation.

This article walks you through setting up the AMFJ MCP server with ChatGPT, Claude, or Claude Code.


What you need

Before you begin, make sure you have:

    • For ChatGPT: a ChatGPT Pro, Plus, Business, Enterprise, or Edu account with Developer mode enabled
    • For Claude: a paid Claude plan (Pro, Max, Team, or Enterprise). Custom connectors are not available on the free plan.
    • For Claude Code: a paid Claude plan (Pro, Max, Team, or Enterprise)

Enabling Developer mode in ChatGPT

For Plus and Pro users:

  1. Go to SettingsSecurity and login
  2. Turn on Developer mode

For Business, Enterprise, and Edu workspaces:

An administrator must enable Developer mode:

  1. Go to Workspace SettingsPermissions and RolesConnected Data
  2. Enable Developer mode
  3. For Enterprise and Edu, you can grant access to specific people or groups

Enterprise and Edu administrators can also publish apps to the entire workspace instead of having each person add them individually. See Publishing to a workspace below.


Set up the AMFJ MCP


Choose the instructions for your AI assistant:

ChatGPT

  1. Open Plugins from the left sidebar in ChatGPT
  2. Click the plus button to start a new plugin
  3. Name it AMFJ (or another name you prefer). The description is optional.
  4. Leave Connection set to Server URL (not Tunnel)
  5. In the URL field, enter:
https://app.assetmanagementforjira.com/mcp
  1. Under Authentication, choose OAuth
  2. Leave the Advanced OAuth settings unchanged
  3. Under the risk warning, tick I understand and want to continue
  4. Click Create
  5. Sign in with your Atlassian account
  6. If prompted, choose your AMFJ site
  7. Review the permissions AMFJ is requesting
  8. Before clicking Connect, verify that the line under the button says access is being sent to chatgpt.com
  9. Click Connect

You'll be redirected back to ChatGPT, and the plugin will be ready to use.

Note: Older guides may mention a "Scan Tools" step, but the personal setup flow does not include one. Clicking Create performs discovery automatically.


Claude

  1. In Claude, open SettingsConnectors
  2. Choose Add custom connector
  3. Enter the server URL:
https://app.assetmanagementforjira.com/mcp
  1. Click Connect
  2. Sign in with your Atlassian account
  3. If your account can access more than one AMFJ workspace, choose the workspace you want to connect
  4. Review the permissions AMFJ is requesting
  5. Before clicking Connect, verify that the line under the button says access is being sent to claude.ai
  6. Click Connect

You'll be redirected back to Claude, and the connector will be ready to use.


Claude Code

  1. Open your terminal
  2. Run the following command:
claude mcp add --transport http amfj https://app.assetmanagementforjira.com/mcp
  1. Claude Code will confirm the connector has been added
  2. Open Claude Code and start a new conversation
  3. Type /mcp  and choose to authenticate
  4. Your browser will open with three screens:
    • Sign in with Atlassian (if you're not already signed in)
    • Atlassian asks to install the app on a site: Choose your AMFJ site and click Accept
    • AMFJ shows its permission screen: Review the permissions, then click Connect
  5. The browser tab will say Authentication successful and can be closed
  6. Claude Code will pick up the result, and the connection will be active

Important: Do not use the --header  flag when adding the AMFJ MCP server. Passing an API key that way skips OAuth and gives the connection a shared tenant-wide credential instead of your own permissions.


Check that it worked

Once you've connected AMFJ, try asking your AI assistant a question about your inventory:

How many assets do we have?

or

What does Sarah have checked out?

The assistant should respond with real data from your AMFJ workspace.

In ChatGPT, you can also try a write action:

Add a MacBook Pro to inventory

ChatGPT will propose the record and wait for your confirmation before creating it in AMFJ.


What the connection can and cannot do

The AMFJ MCP server acts with your own AMFJ permissions:

  • Permissions are resolved on every request. If you have read-only access in AMFJ, ChatGPT or Claude can search and view your inventory but cannot check assets in or out, create records, or make changes.
  • Access lasts an hour at a time and renews automatically for up to 30 days of continued use.

Publishing to a workspace

This section applies only to ChatGPT Business, Enterprise, and Edu administrators.

When you create the AMFJ plugin as described above, it starts as a draft marked Dev. Workspace members cannot see or use it until you publish it.

To publish the plugin:

  1. Go to your workspace app settings
  2. Find the AMFJ plugin
  3. Publish it to your workspace

For Enterprise and Edu, you can choose which actions are allowed and which groups get access.

Once published, members can use the AMFJ plugin in their ChatGPT conversations. Each member will still need to authorize the connection with their own Atlassian account the first time they use it.


Troubleshooting

The OAuth popup doesn't open, or I'm not redirected back after signing in

  • Make sure pop-ups are enabled for chatgpt.com  or claude.ai  in your browser
  • If you're using a browser extension that blocks redirects, try disabling it temporarily
  • Try opening the AI assistant in a different browser or incognito window
  • For ChatGPT, make sure you've clicked Connect after signing in. Adding and connecting are separate steps.

I don't see the AMFJ connector (workspace members)

  • In ChatGPT Business, Enterprise, or Edu: Check with your workspace administrator to confirm the plugin has been published and you've been granted access
  • In Claude Business: An administrator may need to add the connector before members can see it

The AI assistant says it can't access my AMFJ data

  • Confirm you've completed the OAuth authorization flow
  • Check that your Atlassian account has access to the AMFJ workspace you're trying to query
Did this answer your question? Thanks for the feedback There was a problem submitting your feedback. Please try again later.

Still need help? Contact Us Contact Us