Rovo MCP tools/call fails with generic error "We are having trouble completing this action. Please t

[Ezhumalai Nagasundaram]

We are integrating Atlassian Rovo MCP (Model Context Protocol) with Jira Cloud via OAuth.

We are experiencing a consistent issue where MCP tool execution fails.

---

Environment:

• Product: Jira Cloud + Rovo MCP
• MCP URL: ROVO MCP endpoint (via Atlassian MCP server)
• Cloud ID: 1dfba9bb-c284-4e1e-bc9d-59349ca0f4a8
• Auth type: OAuth 2.0 (3LO)
• Scopes used:
  • read:jira-work
  • write:jira-work
  • read:issue:jira
  • offline_access
  • read:account

---

Steps to reproduce:

1. Perform OAuth authentication and obtain access token + refresh token
2. Initialize MCP session:
   • method: initialize
   • returns valid MCP session ID ✔
3. Call tools/list
   • returns available tools successfully ✔
4. Call tool:
   • searchJiraIssuesUsingJql
   • with valid arguments:
   • Not only for this tool, failing for all the tools.

{
  "cloudId": "1dfba9bb-c284-4e1e-bc9d-59349ca0f4a8",
  "jql": "project = KAN ORDER BY updated DESC",
  "maxResults": 20
}

Actual result:

tools/call fails with:

{
  "result": {
    "content": [
      {
        "type": "text",
        "text": "{\"error\":true,\"message\":\"We are having trouble completing this action. Please try again shortly.\"}"
      }
    ],
    "isError": true
  }
}

Expected result:

• MCP tool should return Jira issue search results
• Same query works correctly via Jira REST API:

POST /rest/api/3/search/jql

Observations:

• MCP initialize succeeds ✔
• MCP tools/list succeeds ✔
• Tool exists in registry ✔
• Arguments are valid ✔
• Jira REST API works correctly ✔
• Only MCP tools/call fails ❌
• Error is generic and non-actionable
• No schema validation error returned

---

Additional context:

We also checked Rovo insights:

• “Users with AI enabled: 1”
• No active Rovo interactions shown
• No usage analytics populated

This suggests MCP/Rovo execution layer may not be fully active for the tenant.

---

Hypothesis:

This appears to be one of the following:

• MCP tool execution not fully enabled for tenant
• Rovo AI execution backend not provisioned
• Partial rollout / feature flag mismatch
• MCP execution service unavailable for this cloud site
• Silent failure in tool execution layer

---

Impact:

• MCP integration cannot be used for Jira automation
• Only REST API fallback is currently usable
• Blocks production usage of Rovo MCP tools

---

Request:

Please confirm:

1. Whether MCP tool execution (tools/call) is enabled for this Jira Cloud tenant
2. Whether Rovo AI execution backend is fully provisioned
3. If there are any known restrictions or rollout limitations for MCP tools on this site
4. Any required additional configuration for enabling tool execution

---

Notes:

We are able to:

• authenticate successfully
• initialize MCP session
• list tools

But tool execution always fails with a generic error response.

[Alexander Nilsson]

Hello Ezhumalai,

the sequence you described, where initialize and tools/list succeed but every tools/call fails while the same query works over REST, narrows this down well, and the exact error string you are hitting is actually documented.

"We are having trouble completing this action. Please try again shortly." is one of three defined end-user errors for the Rovo MCP server. Per the official docs it maps to a transient backend error that happens during the permission check, not a malformed request or an OAuth scope problem on your side. That lines up with your symptoms: the handshake and the tool registry are healthy, and the failure is server-side at execution time.

One distinction worth making before you conclude the tenant is not provisioned. If a permission were actually blocked at the org level, the server returns a different message: "Access denied: Your organization admin has not authorized the permission." You are not getting that one, so a hard permission block is not the match here. It is still worth a quick check, because the gatekeeper changed recently. The Permissions tab under Atlassian Administration > Rovo > Rovo MCP server now controls Read, Write, and Search, and it takes precedence over Connected Apps and individual Marketplace app permissions. Have your org admin confirm Read and Search are set to Allowed for Jira there.

The error mapping and the Permissions tab are both covered here:

https://support.atlassian.com/security-and-access-policies/docs/Configure-Atlassian-Rovo-MCP-server-permission/

Since the failure is consistent across every tool and has not cleared on retry, you are past the transient case the docs describe, so the documented next step is a support ticket. Send exactly the evidence you already gathered: the same JQL works on POST /rest/api/3/search/jql but fails on tools/call for searchJiraIssuesUsingJql, plus your cloudId and the MCP session details. That is what lets them inspect the execution layer for your specific site rather than treating it as a transient blip.

https://support.atlassian.com/contact/#/

Looking forward to your feedback :) 

Greetings,

Alex