What the Copilot Jira Connector Can and Cannot Read
Short answer: Microsoft 365 Copilot can read Jira through a connector, but only if an administrator deploys it and you hold the Microsoft 365 Copilot add-on license. On a standard Microsoft 365 license, Jira issues show up in Microsoft Search and never in Copilot’s answers. (A Copilot Studio license or pay-as-you-go billing lets agents use them, but still not Copilot Chat; details in the license table below.)
So if your connector is set up and Copilot still answers as if your backlog does not exist, nothing may be broken. Three things decide whether it works: which Jira you run, which license you hold, and what the connector copies across. This article goes through each one, based on Microsoft’s documentation as of 8 October 2026. Microsoft updates these pages often, so check the linked sources before you act on a detail.
Key Takeaways
- Microsoft has two Jira connectors: one for Jira Cloud and one for Jira Data Center. Jira Server has none.
- Indexing is free. Answering from the index needs the Microsoft 365 Copilot add-on (or Microsoft 365 E7).
- Comments: the Jira Cloud connector does not index them, and Microsoft documents no option to add them. On Data Center, comments and attachments come only on request via Microsoft Support.
- With permission mirroring on, the Data Center connector denies access to anything it cannot evaluate. For Jira Cloud, Microsoft documents one unsupported permission type, not a general rule.
- Data arrives on a schedule (Data Center default: every 15 minutes, full crawl daily), so the latest edits can be missing.
Microsoft calls these integrations Microsoft 365 Copilot connectors. In older documentation and in parts of the setup, such as the connector agent and the Atlassian Marketplace plugin, the same technology appears as Microsoft Graph connectors.
Which Jira Versions Does the Copilot Connector Support?
Jira Cloud and Jira Data Center, through two different connectors. Jira Server is not supported.
| Jira Cloud connector | Jira Data Center connector | |
|---|---|---|
| Supported Jira | Jira Cloud only | Jira Data Center (Microsoft’s plugin supports 8.10.0 to 10.5.1) |
| Jira Server | Not supported | Not supported |
| Extra infrastructure | None | Microsoft Graph connector agent on a Windows machine in the same network, version 3.1.15.0 or later |
| Extra software in Jira | None | Microsoft’s plugin from the Atlassian Marketplace |
| Comments and attachments | Not listed as indexed | Only on request via Microsoft Support |
Check one detail before you plan an upgrade. Microsoft’s overview page for the Data Center connector says it supports version 8.14 and later; the deployment page says the plugin supports 8.10.0 to 10.5.1. Taken together, 8.14 to 10.5.1 is the range both pages agree on. On Jira Data Center 11, check the Marketplace listing first.
Which License Does Copilot Need to Answer From Jira?
The Microsoft 365 Copilot add-on, or Microsoft 365 E7. Indexing Jira with a Microsoft-built connector costs nothing extra on any Microsoft 365 tenant; using that index in Copilot’s answers does not come with the base license.
| License | Jira in Microsoft Search | Jira in Copilot answers and agents |
|---|---|---|
| Microsoft 365, any plan | Yes | No |
| Microsoft 365 + Microsoft 365 Copilot add-on | Yes | Yes |
| Microsoft 365 E7 | Yes | Yes |
| Microsoft 365 + Copilot Studio license | Yes | Agents only, no Copilot grounding |
| Microsoft 365 Copilot pay-as-you-go | Yes | Agents only, no Copilot grounding |
That is how a team ends up with a working connector, Jira issues in Microsoft Search, and an empty answer in Copilot Chat.
Deployment needs two kinds of admin rights: the AI administrator role in the Microsoft 365 admin center, and admin access on the Jira side.
Not sure which Copilot license you are on?
The free guide for product owners has the tier table, plus prompts that work on any license, because you supply the content. The download opens straight away and we do not email you the file.
Does the Copilot Jira Connector Index Comments?
On Jira Cloud, no. On Jira Data Center, only on request. For the Jira Cloud connector, Microsoft lists these indexed fields: title, issue description, status, project name, created and updated dates, assignee and priority. You can add custom fields; if an issue type lacks a selected field, it is stored as empty. Comments are not on the list, and Microsoft documents no setting to add them.
That gap matters more than it looks. In most teams the comments hold the actual decision: the developer’s clarification, the edge case someone raised, the reason the scope changed. A Copilot that has read the description but not the thread summarizes the ticket as it was written, not as it was agreed.
The Data Center connector indexes issues (summary, description, priority, status, assignee, project, timestamps) and project metadata. Microsoft can switch on comments and attachments, but only after a request to Microsoft Support.
Does the Jira Connector Respect Jira Permissions?
Yes, if you choose the restricted mode. Both connectors offer two settings:
- Everyone: indexed issues are visible to every user in the tenant.
- Only people with access to this data source: Jira’s permissions are mirrored. Use this for anything confidential.
Mirroring matches Jira users to Microsoft Entra ID accounts by email address. Where the addresses differ, an administrator has to add a mapping rule; without it, people may miss issues they are allowed to see in Jira.
Some permission types cannot be evaluated at all. The Jira Cloud connector does not support the “Any user logged in” application role for granting issue access. The Data Center connector cannot evaluate “Any logged-in user”, “Application access” or some custom-field-based permission types, and denies access in those cases. For the Cloud connector, Microsoft does not document what happens with other permission types it cannot resolve, so test a confidential project before relying on it. On Data Center, the deny rule is the safe default and a frequent reason why one project returns nothing while others work. If that happens, check that project’s permission scheme first.
How Up to Date Is Jira Data in Copilot?
It lags behind Jira by design. For Jira Data Center, Microsoft’s default is an incremental crawl every 15 minutes and a full crawl once a day; both intervals can be changed. For the Jira Cloud connector, Microsoft’s overview page (as of 8 October 2026) documents incremental and full crawls but no default interval.
Permissions can lag further than content. For the Confluence Cloud connector, Microsoft states that permission changes are picked up only in the daily full crawl, not in the 15-minute incremental syncs. Assume the same for Jira until you have tested it.
In practice: don’t ask Copilot about the refinement session you finished ten minutes ago.
What Should a Product Team Check First?
Four things, in this order:
- Which Jira you run. Cloud, Data Center, or Server. Server has no connector.
- Which license you hold. Without the Microsoft 365 Copilot add-on (or E7), Copilot does not answer from Jira, connector or not. Your Microsoft 365 administrator can see which licenses are assigned to you.
- What is indexed. If comments are not, expect summaries of the ticket as written, not as discussed.
- One real test. Pick a project you know well and ask something only a comment or a recent edit can answer, for example “What did we agree about the export format in ABC-123?”. The answer shows the gaps faster than any documentation.
Until all four line up, the method that works is the one most teams already use: paste the item in and give the model the context yourself. Our free guide for product owners has prompts written for exactly that.
When Is the Connector Not Enough?
The connector fits when you run Jira Cloud or a supported Data Center version, hold Copilot add-on licenses, and mainly need to find and summarize issues.
It fits less well when:
- the answer lives in comments,
- you run Jira Server or an older Data Center version,
- the knowledge you need is spread across Jira, Confluence and other systems,
- the index should not live in Microsoft 365 at all.
The alternative is a retrieval system you control. It reads Jira and Confluence through their APIs, keeps the permissions people already have, cites every answer back to the issue or page it came from, and runs on infrastructure you choose. That is what we build at Ironum as enterprise RAG.
Hit one of these limits?
Write to us with your Jira version and what you need Copilot to answer. We reply within two working days with whether it is solvable and roughly what it would take. No sales sequence follows.
Sources
Every statement above comes from Microsoft’s own documentation, checked on 8 October 2026:
- Jira Cloud connector overview: supported deployments, indexed fields, crawl types, permission modes and the “Any user logged in” limitation.
- Jira Data Center connector overview: indexed content, permission hierarchy, the version 8.14 statement and the comments and attachments note.
- Deploy the Jira Data Center connector: connector agent and plugin requirements, the 8.10.0 to 10.5.1 range, and the default crawl schedule.
- Prerequisites for deploying connectors: admin roles, free indexing, and the licensing table.
- Confluence Cloud connector overview: permission-update latency.