nav[aria-label="Primary Navigation"] { padding: 0; & ul { list-style: none; width: 100%; display: flex; flex-direction: row; justify-content: start; align-items: start; gap: 30px; padding: 0; & li { margin: 0; } & ul li { list-style: none; } } }

Rovo MCP server issue

Deepak Chikane

I am also getting the same error " We are having trouble completing this action. Please try again shortly. "

My mcp tool is setup correctly ,tools are also rendered correctly but when I am querying "mcpTools.atlassianUserInfo" it is giving me this error from long time .

Can anyone help me for the same .

Accepted answers

Dr Valeri Colon (Connect Centric)

The responses are heading in the right direction @Deepak Chikane Based on your tests, this doesn’t look like a schema or setup issue anymore. Since all tools load but fail at execution, it points to either auth/context issues (token format, scopes, admin settings) or a platform-side MCP runtime issue. The “Cannot read properties of undefined” especially suggests a backend error. At this stage, if auth is correct, this likely needs Atlassian support to investigate.

All comments

Prasanna Ravichandran

Hi,

Try these quick checks first:

Confirm the MCP server is connected and the tool schema is loading correctly, since you mentioned the tools are rendering.
Test with another MCP tool if available, just to verify whether the issue is only with mcpTools.atlassianUserInfo or with MCP execution in general.
Recheck the tool input/output schema for mcpTools.atlassianUserInfo, because sometimes the tool appears in Rovo but fails at runtime if the schema or response format is not exactly what Rovo expects.
Review the MCP server logs while triggering the action. That usually shows whether the request is reaching the server or failing before execution.
Try reconnecting the MCP integration or re-registering the tool, as stale registration can sometimes cause this generic “Please try again shortly” error.

Since the error is happening specifically for mcpTools.atlassianUserInfo and not during tool rendering, this looks more like a runtime execution or response-format issue rather than a setup issue.

Himanshu Tiwary_go407r

HI @Deepak Chikane ,

It sounds like your MCP server connection is only working partially.

If the tools are showing up correctly, that usually means the server registration and tool discovery are fine. The failure is more likely happening when the tool is actually invoked. Since mcpTools.atlassianUserInfo is a user-context tool, this kind of error often points to one of these areas:

the tool is missing the required auth scope or user context

the server is returning a response format Rovo does not like

the tool call is timing out or failing internally even though it renders successfully

there is a platform-side issue with that specific tool/action rather than with the whole MCP setup

A few things I would check:

Test with a very simple custom MCP tool

If a basic tool works but mcpTools.atlassianUserInfo fails every time, that helps narrow it down to that tool specifically.

Check the MCP server logs during invocation

See whether the request is actually reaching your server. If it is, check the response body, status, and timing. If it is not, the problem may be on the Rovo side.

Validate the response schema carefully

Even when a tool renders in Rovo, invocation can still fail if the returned payload does not exactly match what the client expects.

Confirm auth and permissions

Since this tool is user-info related, make sure the requesting user context is available and that the app has the right scopes/permissions.

Try another tool in the same server

If other tools execute successfully, that strongly suggests the issue is isolated to atlassianUserInfo rather than your full MCP server setup.

Re-check for timeout or long-running logic

Sometimes the generic “Please try again shortly” message hides a timeout or backend exception.

If you have already verified all of the above, then this may be a product-side issue rather than a configuration mistake on your end, especially if it has been failing consistently for a long time with the tool otherwise visible and discoverable.

It would help if you could share:

whether the request reaches your MCP server

any server-side error logs

the response shape returned by mcpTools.atlassianUserInfo

whether other MCP tools execute successfully

That would make it much easier to tell whether this is a schema, auth, timeout, or Rovo platform issue.

If this helped, please give it an upvote.

Deepak Chikane

Hi @Prasanna Ravichandran ,
Thank you for the troubleshooting suggestions. I've completed all the diagnostic checks you recommended. Here are the findings.

1. MCP Server Connection & Tool Schema Loading
Status: WORKING
MCP API URL: https://mcp.atlassian.com/v1/mcp
CloudId:
Token exists: true
Site URL: https://gravie.atlassian.net
MCP client created: SUCCESS
Tools loaded: 20+ tools

All tool schemas load correctly.

2. MCP Tool Execution Tests

Status: All tools failing

Test 1: atlassianUserInfo

Input:
{}

Result:
{
content: [...],
isError: true
}

Error message:
"We are having trouble completing this action. Please try again shortly."

Test 2: searchJiraIssuesUsingJql

Input:
{
"cloudId": "xyz",
"jql": "assignee = currentUser() ORDER BY updated DESC"
}

Result:
Error message:
"We are having trouble completing this action. Please try again shortly."

Test 3: search

Input:
{
"query": "test query"
}

Result:
{
content: [...],
isError: true
}
Error:Cannot read properties of undefined

3. Schema Verification:
Status: Correct
atlassianUserInfo → Called with {}
searchJiraIssuesUsingJql → cloudId + jql provided
search → query provided
All parameters match the tool schemas.

Deepak Chikane

Hi @Himanshu Tiwary Thank you for the troubleshooting suggestions. I've completed all the diagnostic checks you recommended. Here are the findings.

1. MCP Server Connection & Tool Schema Loading
Status: WORKING
MCP API URL: https://mcp.atlassian.com/v1/mcp
CloudId:
Token exists: true
Site URL: https://gravie.atlassian.net
MCP client created: SUCCESS
Tools loaded: 20+ tools

All tool schemas load correctly.

2. MCP Tool Execution Tests

Status: All tools failing

Test 1: atlassianUserInfo

Input:
{}

Result:
{
content: [...],
isError: true
}

Error message:
"We are having trouble completing this action. Please try again shortly."

Test 2: searchJiraIssuesUsingJql

Input:
{
"cloudId": "xyz",
"jql": "assignee = currentUser() ORDER BY updated DESC"
}

Result:
Error message:
"We are having trouble completing this action. Please try again shortly."

Test 3: search

Input:
{
"query": "test query"
}

Deepak Chikane

Hi ,
Thank you for the troubleshooting suggestions. I've completed all the diagnostic checks you recommended. Here are the findings.

1. MCP Server Connection & Tool Schema Loading
Status: WORKING
MCP API URL: https://mcp.atlassian.com/v1/mcp
CloudId:
Token exists: true
Site URL: https://gravie.atlassian.net
MCP client created: SUCCESS
Tools loaded: 20+ tools

All tool schemas load correctly.

2. MCP Tool Execution Tests

Status: All tools failing

Test 1: atlassianUserInfo

Input:
{}

Result:
{
content: [...],
isError: true
}

Error message:
"We are having trouble completing this action. Please try again shortly."

Test 2: searchJiraIssuesUsingJql

Input:
{
"cloudId": "",
"jql": "assignee = currentUser() ORDER BY updated DESC"
}

Result:
Error message:
"We are having trouble completing this action. Please try again shortly."

Test 3: search

Input:
{
"query": "test query"
}

Deepak Chikane

Status: All tools failing

Test 1: atlassianUserInfo

Input:
{}

Result:
{
content: [...],
isError: true
}

Error message:
"We are having trouble completing this action. Please try again shortly."

Test 2: searchJiraIssuesUsingJql

Input:
{
"cloudId": "",
"jql": "assignee = currentUser() ORDER BY updated DESC"
}

Result:
Error message:
"We are having trouble completing this action. Please try again shortly."

Test 3: search

Input:
{
"query": "test query"
}

Himanshu Tiwary_go407r

Hi @Deepak Chikane ,

Thanks for sharing the detailed test results. Based on what you posted, this does not look like a tool schema problem anymore. Since the MCP server connects, the tools render, and all executions fail at runtime, the issue is more likely related to authentication, scopes, or org/admin policy during invocation rather than MCP registration itself. Atlassian’s troubleshooting guide points to exactly those areas when setup looks correct but tool execution still fails.

A few things stand out from your examples:

atlassianUserInfo depends on user identity access

searchJiraIssuesUsingJql needs Jira read permissions

search is a separate Rovo search capability

Atlassian documents these tool requirements in the supported tools list, so it is possible for tools to appear correctly in discovery but still fail when the granted auth context is incomplete at execution time.

The first thing I would verify is the auth method you are using. If this is a personal Atlassian API token, Atlassian says it must be sent via HTTP Basic Auth using Basic base64(email:api_token). If you are sending that personal token as Bearer, the tools can fail even though the MCP endpoint itself is reachable.

I would also check whether API token authentication is enabled by your organization admin. Atlassian says API-token auth is optional and must be allowed in the Rovo MCP server settings. If that setting is off, execution can fail even when discovery works.

Another thing to confirm is whether your org/site controls are allowing the request path fully. Atlassian notes that access can also be affected by admin-managed controls around Rovo MCP usage and supported/custom domains, so this can show up as a runtime failure rather than a schema error.

Given your results, my conclusion is:

not a schema-loading issue

likely an auth/scopes/admin-policy issue

possibly a product-side runtime issue for search, because Cannot read properties of undefined looks more like an internal execution bug than an input mismatch

The cleanest retest order would be:

re-authenticate from scratch

verify correct auth format

test atlassianUserInfo

test getAccessibleAtlassianResources

then test searchJiraIssuesUsingJql with the returned cloudId

test search last

If all of those still fail after re-auth with the correct method and permissions, I would treat it as an Atlassian-side MCP runtime issue and raise it with Atlassian Support, because at that point your configuration looks structurally correct and the failure is happening during execution.

Dr Valeri Colon (Connect Centric)

Sumit Mahajan_52xsnz

Im facing same issue @Deepak Chikane Have you got any solution for the same ?

Welcome!

It looks like you're new here. Sign in or register to get started.

Quick Links

Find more posts tagged with

rovo-atlassian-intelligence