Atlassian’s remote MCP is hosted at mcp.atlassian.com/v1/sse (SSE) in many clients and works with a normal Atlassian account. You authenticate with an Atlassian API token and your product email, then give the model the same Jira and Confluence reach you have through the product UI — so it can file issues, find pages, and connect specs to epics in one pass.
https://mcp.atlassian.com/v1/mcp
Claude Sonnet 4.5
Chat with 60+ AI models on the same workflow — switch to a different model mid-conversation and re-run the same prompt, or use Compare mode to put several side-by-side and balance quality vs. cost.
OAuth 2.1 through your Atlassian account, or an API token from id.atlassian.com (Security → API tokens). Tools run with your existing permissions.
How models use it and what it is built for.
The Jira and Confluence MCP is aimed at product and engineering orgs: search and read Confluence runbooks while updating Jira epics, create sub-tasks from a meeting note, and cross-link documentation to active bugs. The server’s tools reflect what your Atlassian Cloud user can do with your existing permissions, so a contributor cannot magically bypass project roles. In MCP Playground, use the Jira and Confluence template: it keeps the right URL and token hints in sync with Atlassian’s documentation.
Typical tools an AI model can call. Exact names vary by version.
Taken from the official Jira & Confluence documentation — see Atlassian Remote MCP Server for the full reference.
Examples
Streamable HTTP (current)
https://mcp.atlassian.com/v1/mcpThe endpoint to use for new setups. Authorises over OAuth 2.1.
SSE (legacy)
https://mcp.atlassian.com/v1/sseStill live, for clients that do not yet speak streamable HTTP. Prefer the endpoint above where you can.
Client configuration
Any MCP client
First connection opens Atlassian OAuth in the browser; the agent then acts within your own permissions.
{
"mcpServers": {
"atlassian": {
"type": "http",
"url": "https://mcp.atlassian.com/v1/mcp"
}
}
}Copy any of these into MCP Agent Studio after connecting.
List all In Progress items assigned to me in the Platform board with blockers called out.
Create a Jira task under the Mobile epic titled “Fix crash on cold start for iOS 18”.
Find the Confluence page that describes our release train and pull the key dates.
This is not a single-model product: you get the same MCP connection with 60+ models (Claude, GPT, Gemini, DeepSeek, open-weight, and more), you can switch mid-conversation, and you can open Compare mode to run the same prompt against multiple models at once. The card above is a suggested starting point for this server — not the only choice.
Default pick for Jira & Confluence
Claude Sonnet 4.5
Claude Sonnet 4.5 is best for long Confluence context plus Jira state changes. For quick JQL lookups, Haiku 4.5 is enough; test both in Agent Studio.
Listing tools proves the server is reachable, not that a model can work with it. Evals go further: they read every tool on the server, write a test suite from its real schemas, and run it — code decides pass/fail on the responses (schema conformance, error codes, pagination, result caps) while a scoring model grades plain-English tasks driven through the tools.
Get a pass/fail report per tool with the evidence behind each verdict — and replay the same suite after every schema change. Destructive tools are excluded from the run.
Run evalsOpen MCP Agent Studio with the connection pre-filled. Add your token, pick any of 60+ models, and start chatting — no install required.
Try Jira & Confluence in Agent StudioCommon questions about connecting, scoping and using it safely.
It is Atlassian’s remote MCP service for cloud products that exposes Jira work items, boards, and Confluence pages to AI tools through a standard tool surface, using your existing Atlassian account permissions rather than a separate shadow integration.
Create a personal Atlassian API token in your Atlassian account, keep your work email, and add both to the MCP client according to the server’s expected scheme — often a paired email and token for HTTP basic auth, or a bearer form your client documents.
The model is limited to what your Jira and Confluence roles allow. A reader cannot close epics, and a Jira Software user cannot see pages in a restricted Confluence space. If you are testing write behaviour, do it in a test project with disposable issues.
The hosted mcp.atlassian.com endpoint is built for Atlassian Cloud. On-prem and Federation setups may require different network paths; consult your admin before routing production traffic through a third-party MCP client.
Claude Sonnet 4.5 is a good default when you are jumping between a long spec page, comments, and several Jira issues, because the model can hold cross-links in context. You can A/B a cheaper model in Agent Studio for pure JQL one-liners.