Hi everyone,
I'm seeing an issue with Rovo when working across Jira and Confluence.
When I'm in Jira, Rovo can successfully:
However, when Rovo attempts to open and read the linked Confluence page, it fails and never returns the page content.
The page:
After this, Rovo enters a loop investigating tools and schemas instead of returning a result.
Has anyone successfully got Rovo to:
Or is this a known limitation/issue with cross-product retrieval?
Thanks
Hello @Carol Pretorius ,
@Mia Tamm _Simpleasyty_ isolation is the right one, and your answer to it (Rovo in Confluence reads the page, Rovo in Jira resolves the link but fails on the content) points at a specific layer that "the page is unrestricted" does not cover, because it sits above page permissions.
The first thing to check: a data security policy with an AI control on that space. In Atlassian Administration, Security, then Data protection, then Data security policies, a policy can carry controls that block Rovo or AI features from reading content in specific spaces, with per-space overrides. Atlassian's own KB describes the exact symptom you have: the space is visible and the page resolves, but the content cannot be read by the AI path, while a user (and Rovo inside Confluence itself) reads it fine (Unable to read Confluence pages within Rovo but can view spaces). The KB is written for the MCP path, but the policy applies to any cross-product read. Ask your org admin to open each policy, look at Overrides on the controls, and see whether that space is listed. If it is, that is your cause, and the test is adding a throwaway space to the same override and repeating the Jira read.
If no policy names the space, two more layers, one test each:
One last thing worth writing in the eventual support ticket, if it comes to that: the exact wording Rovo returns when it fails ("I don't have access," "I couldn't retrieve," or a generic error) tells support which layer refused, and they will ask for it first.
This is super helpful, thanks @Sami Shaik, especially the data security policy angle. That gives Carol a much cleaner way to isolate whether the failure is permissions, site context or Rovo access itself.
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
Thanks @Mia Tamm _Simpleasyty_ , one refinement for @Carol Pretorius , prompted by re-reading the original post: the error Rovo returns is a 404, "The requested resource could not be found." That matters for the order of the checks. A data security policy blocking an AI read presents as a refusal (no access, cannot retrieve), not as not-found. A 404 on a page that exists and is readable is the signature of Rovo resolving the page against the wrong context: the Confluence site it is reading from inside Jira is not the one the page lives on, or the page reference it obtained from the link does not resolve in that context.
So for Carol I would run the site-context test first: link a page from a Confluence space on the same site as the Jira space and ask Rovo in Jira to read it; if that works and the original page (on another site, or reached through a Smart Link to a different site) still 404s, that is the cause, and it is a product limitation to report with the two page URLs rather than a policy to change. The data security policy check stays second, for the case where the error wording changes to a refusal. And the loop through tools and schemas after the 404 is Rovo retrying a read that will never succeed, which is consistent with a resolution failure rather than a permission one.
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
Good catch — the 404 changes the order of operations quite a bit.
Testing same-site vs cross-site first makes much more sense here. Thanks for the refinement, @Sami Shaik.
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
hi @Mia Tamm _Simpleasyty_ yes Rovo in Confluence fine, this happens from Rovo in Jira trying to read content on an unrestricted Confluence page.
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
Hi @Carol Pretorius interesting that Rovo can resolve the linked Confluence page but fails only when it tries to read the content.
I’d test the same page directly from Rovo in Confluence, then repeat with a brand-new unrestricted page in the same space. If both work directly but fail only when Rovo follows the Jira → Confluence link, you’ve got a much cleaner reproduction of a cross-product retrieval issue rather than a page permission problem.
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
Hi @Mia Tamm _Simpleasyty_ yes Rovo in Confluence fine, this happens from Rovo in Jira trying to read content on an unrestricted Confluence page.
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.