Hello everyone,
I’ve created a Rovo Agent in Atlassian Studio and integrated it into our Slack channel. While the native Jira and Confluence skills are working perfectly, I’m having trouble with the GitHub MCP (Model Context Protocol) integration.
The Goal:
I want the Rovo Agent to be able to retrieve logs and status information from GitHub Actions pipeline links or specific repositories mentioned in Slack.
The Problem:
Even though I have enabled the GitHub MCP server and all its associated skills in the Agent configuration, every time I ask a question that requires GitHub access, Rovo responds with:
"This response was generated without using the following MCP server(s), because you're currently not authenticated. Your response quality may improve if you authenticate to these servers. Select the server(s) you'd like to authenticate to:"
What I’ve Tried:
Verified the GitHub MCP is enabled in the Agent's "Skills" section.
Ensured the Agent has the correct instructions to use GitHub tools.
Checked that my Atlassian account is linked, but the prompt for GitHub authentication keeps appearing.
Questions:
Is there a specific step to perform a "Service Principal" or "Admin-level" authentication for the GitHub MCP so the Agent doesn't require individual user auth every time?
Are there specific OAuth scopes or GitHub Organization settings (like third-party application restrictions) that might be blocking Rovo from maintaining a session?
How can I ensure the Agent uses the authenticated connection to parse pipeline links?
Any guidance on how to fix this authentication loop would be greatly appreciated!