atlassian-mcp-server vs mcp-atlassian
atlassian-mcp-server: The Atlassian Rovo MCP Server acts as a secure, cloud-hosted bridge connecting AI tools to various Atlassian products like Jira, Confluence, and Bitbucket. It enables real-time data access and interaction, allowing AI agents to summarize, search, create, and update content within these platforms while respecting user permissions.; mcp-atlassian: MCP Atlassian is a Model Context Protocol server designed for seamless integration with Atlassian's Confluence and Jira products. It enables AI assistants to interact with both Cloud and Server/Data Center deployments of these platforms.
mcp-atlassian is a community server you run yourself, covering Jira and Confluence on Cloud and Server/Data Center. The official one is hosted and Cloud only, with more products.
TL;DR
- Your team is on Atlassian Cloud and you want nothing to install: clients connect to Atlassian's hosted endpoint.
- You want OAuth 2.1 sign-in where the agent acts with each user's existing permissions and IP allowlists apply.
- You need more than Jira and Confluence: Jira Service Management, Bitbucket and Compass are covered too.
- You run Jira or Confluence on Server or Data Center: the service is Atlassian Cloud only.
- You need to self-host or audit the server code: the repository holds documentation, not the server source.
- Your data must not pass through a hosted service: requests go through Atlassian's endpoint.
- You run Jira or Confluence on Server or Data Center (Jira 8.14+, Confluence 6.0+), which the hosted official server does not cover.
- You want to run the server yourself with `uvx mcp-atlassian` and keep it on your own machine or network.
- You want JQL and CQL search, issue transitions and an SLA calculation tool in one MIT-licensed server.
- You need Bitbucket, Jira Service Management or Compass: the README covers Jira and Confluence.
- You do not want API tokens stored in your MCP client config: this server reads them from environment variables.
- You are on Python 3.14: the README says it is not supported yet.
Side-by-Side Comparison
Features
Use Cases
Best For
FAQ
Deploy on DigitalOcean
Host your MCP servers, Docker containers, and AI agents with $200 free credit over 60 days.