Connect Jira Service Management
Jira Service Management is Atlassian's service desk. It holds the customer requests your team answers, the queues they work from, the customers and organizations behind them, the knowledge base, the…
Written By Christopher Scaminaci
Last updated About 3 hours ago
Jira Service Management is Atlassian's service desk. It holds the customer requests your team answers, the queues they work from, the customers and organizations behind them, the knowledge base, the on-call schedules and alerts, the Assets configuration database, and, because it is built on Jira, the issues, projects, workflows, fields and permissions underneath all of it. StackJack talks to it through Atlassian's own REST APIs, the same data you see when you sign in, reachable by your AI assistant.
Connecting Jira Service Management to StackJack gives your AI assistant a large family of jsm_ MCP tools. MCP (Model Context Protocol) tools are the standardized commands an AI assistant can call through StackJack. With them, your AI can:
- Work the service desk - find and read customer requests, add comments, answer approvals, add participants, see SLA clocks, move a request through its workflow, and read the request types, queues, customers and organizations of each service desk
- Search and manage Jira issues - run a JQL search, read or create issues, change fields, assign, transition, log work, link and watch issues, and handle comments, attachments, votes and bulk operations
- Administer projects and people - projects, versions, components, roles and role members, categories, users, groups and avatars
- Shape how Jira behaves - custom fields with their contexts and options, screens, issue types, statuses, priorities and resolutions
- Manage schemes, workflows and site settings - workflows and workflow schemes, permission, notification and issue-security schemes, dashboards, filters, plans, audit records and site configuration
- Run operations and on-call - alerts (acknowledge, assign, escalate, close, add notes), on-call schedules, rotations and overrides, escalation and routing rules, teams, integrations, heartbeats, maintenance windows and audit logs
- Query Assets - objects, object types, schemas and attributes with AQL, and import configuration
- Bring files in and out - attachments, avatars, icons and on-call calendars as temporary download links, and file uploads to issues, service desks and avatars
How StackJack connects to Jira Service Management
You sign in. There is nothing to register, no API token to create and nothing to paste.
Open the Jira Service Management connector in StackJack, click Authorize, sign in to Atlassian, choose the Jira site you want to connect and accept the permissions StackJack asks for. That is the whole setup. StackJack keeps the connection current afterwards: you do not sign in again unless you remove StackJack's access in your Atlassian account settings or leave the connection unused for a long time.
Atlassian does not allow a third party to store personal API tokens, so signing in is the only way to connect, and it is also why you never see a key to copy.
One connection is one Jira site
An authorization covers exactly one Jira site. There is no site switcher and no view across sites. If you look after several customers' Jira sites, you add one connection per site and authorize each separately:
- Click Add connection.
- Name it after the customer or the site, for example "Acme Jira", not "Jira 2".
- Authorize, and choose that site on Atlassian's screen.
Your AI then knows which site you mean: "show me the open requests for Acme" reaches the Acme connection. Ask for the list of connections if you are not sure which sites are set up.
Steps
- Open the Jira Service Management connector in StackJack and click Add connection (or Configure for your first one).
- Name the connection after the site it reaches.
- Click Authorize. Sign in to Atlassian, choose the Jira site and accept the permissions.
- Run Test Connection. It lists one service desk, which confirms the sign-in, the site, that Jira Service Management is on it and that the account is a service desk agent.
- Repeat for each additional site.
A notice on Atlassian's screen that the app has not been reviewed yet is expected while the review is in progress. It does not change what the connection can do.
Who to sign in as
StackJack acts as the Atlassian user who signed in. It can see and change exactly what that person can, and nothing more:
- To work requests, sign in as a service desk agent on the service desks you want your AI to reach. Customer-portal-only users cannot use most of the service desk tools.
- To use the administration tools (fields, screens, workflows, schemes, permissions, users and site settings), sign in as a Jira administrator.
- To use the Assets tools, the account needs access to the site's Assets workspace. Assets is included in Jira Service Management Standard, Premium and Enterprise, each with its own free object allowance.
- To use the operations tools (alerts and on-call), the account needs the operations role that grants what you ask it to do, for example team admin to change a team's schedule.
If a tool answers that the user does not have permission, the fix is in Jira: sign in as someone who can do it by hand, then authorize the connection again.
Who in your organization can use the connection
The connection is authorized once and belongs to your organization, but it carries the Atlassian identity of the one person who signed in. For that reason StackJack runs Jira Service Management tools only for workspace owners: a call from any other member is refused rather than run as the owner, and a member cannot connect a personal sign-in of their own. Assigning Jira tools to a member who is not an owner therefore does not give that member access.
What the connection can do inside Jira
Atlassian gives StackJack its whole permission list in one sign-in, because asking again later would mean every connection signing in again. What each tool can then do is decided in two places:
- Which tools you assign. Assign only the read tools and your AI can look but not touch.
- Which plan the connector is on. Reads are on the Free tier and every write needs Pro. Changes that cannot be taken back through StackJack, or that reach a person, are additionally marked destructive. Business offers the same tool set as Pro with a higher monthly call quota.
See the generated Jira Service Management tool reference for the current inventory, plan assignment, input schemas, and destructive-action labels.
Things worth knowing before you use the write tools
A public comment emails the customer
On a service desk request, a comment is visible to the customer and emails them unless it is marked internal. StackJack's comment tools default to internal: if your instruction does not say otherwise, the comment stays inside your team. The same default applies when a comment travels inside another change, such as an issue update, a workflow transition, a bulk create or an issue link. To reach the customer, say so explicitly ("reply to the customer", "make it public"). Because a public comment reaches a real person, the comment tools are marked destructive, so your AI application can ask you to confirm. Whether it does depends on that application's own settings; see Destructive tools and confirmation.
The same care applies to inviting customers, answering an approval and revoking a customer's portal access: each of them changes something a customer or a workflow acts on, and each is marked destructive.
Jira text is a structured format
Jira Cloud stores descriptions and comments in Atlassian Document Format, a JSON structure, not plain text. Your AI knows to send that format for descriptions and comments. If you paste raw JSON into a request yourself, use the same format.
Deleting cannot be undone
Deleting an issue, comment, attachment, version, component, custom field, workflow or scheme is permanent through StackJack. Every delete is marked destructive. Where Jira offers a softer alternative, such as archiving a project or moving it to the trash, prefer that.
Administration changes affect everyone on the site
Changing a workflow, a permission scheme, a field configuration or a screen changes how every project that uses it behaves, immediately. Review the list of tools you allow before you turn the administration areas on, and try changes on a test project first.
Bulk operations run in the background
Bulk edit, bulk delete, bulk transition and project deletion are asynchronous in Jira. The call returns a task to follow, and the tool descriptions say how to read its progress. Check the result before you rely on it.
Files: attachments, avatars and calendars
Files come out of Jira Service Management as a temporary link rather than the file itself, because an AI assistant's tool results carry text. This applies to request and issue attachments and their thumbnails, avatars, Assets icons and the on-call calendar exports. The link is good for about half an hour, and files over 64 MB are refused.
Any link lets whoever holds it read the file for as long as it lasts, so treat one like a password: use it, do not forward it, and ask again if it has expired. Nothing keeps a second copy.
Going the other way, you can attach a file to an issue, add a temporary attachment to a service desk, or set an avatar. Give your assistant the file directly, or a public https link it can fetch, up to 25 MB.
Searching, paging and size
- Issue search uses Atlassian's current JQL search, which pages with a token rather than a page number. Ask for a narrow query ("open requests in project ACME updated this week") rather than everything.
- Service desk lists page by start position, and a short page is not always the last one; the tool descriptions say how to tell.
- Page size is capped at 100 by StackJack. Ask your AI to narrow by project, status or date rather than reading every page.
- Responses can be large. If a result is too big to return, narrow the request.
Rate limits
Atlassian meters its API by a points budget, and StackJack's Atlassian app has one hourly budget shared by every StackJack customer. In a rare busy hour that budget can run out for everyone at once. When it does, every Jira Service Management tool stops until the top of the next UTC hour and answers that the quota is used up. Nothing is wrong with your connection; try again after the hour rolls over. Separately, Atlassian limits how fast one site can be called, and a retry after a short pause is the remedy.
Troubleshooting
- "not been authorized yet" - the connection has no sign-in. Open the connector and click Authorize.
- "no longer accepts this connection's sign-in" - the sign-in was removed, replaced or left unused for 90 days. Click Authorize again.
- "authorized without the permission this tool needs" - click Authorize again so Atlassian can show the full list of permissions.
- "refused this operation" - the Jira user who authorized the connection cannot do that in Jira. Sign in as a user who can, then authorize again.
- "no Assets workspace" - the site has no Assets workspace the signed-in user can reach. Check the plan and the user's Assets access.
- A different site than you expected - an authorization is one site. Add a connection for the other site.
Jira Service Management tools
jsm_ · 867 tools · Free 421 · Pro 446
Assets workspace
Configuration
Customers
Icons
Imports
Knowledge base
Object types
Objects
Organizations
Request types
Requests
Schemas
Service desk info
Service desks
Issue attachments
Issue bulk operations
Issue comments
Issue link types
Issue links
Issue properties
Issue remote links
Issue search and JQL
Issue votes
Issue watchers
Issue worklogs
Issues
Avatars
Groups
Licenses
Permissions
Project categories
Project classification levels
Project components
Project permission schemes
Project properties
Project role actors
Project roles
Project versions
Projects
Users
Custom fields: contexts and options
Field schemes
Issue field configurations
Issue fields
Issue type schemes
Issue type screen schemes
Issue types
Priorities
Priority schemes
Resolutions
Screen schemes
Screen tabs
Screens
Statuses
Audit and tasks
Dashboards
Filters
Issue notification schemes
Issue security schemes
Permission schemes
Plans
Site settings
Workflow schemes
Workflows
Alerts
Audit logs
Contacts
Custom user roles
Escalations
Forwarding rules
Heartbeats
Integrations
Jira Edge Connector
Maintenances
Notification rules
On-call
Policies
Routing rules
Schedules
Syncs
Team policies
Team roles
Teams
Was this helpful?
More in Connector guides
Connect Acronis Cyber Protect CloudConnect Action1Connect AddigyConnect AlertOpsStill need help? Ask the team