What is Atlassian MCP Server (Jira & Confluence)?
Atlassian MCP Server (Jira & Confluence) is the community mcp-atlassian project: an open-source MCP server that lets AI assistants search, read, create and update Jira issues and Confluence pages on both Atlassian Cloud and Server/Data Center.
It is maintained by sooperset on GitHub (sooperset/mcp-atlassian), published on PyPI as mcp-atlassian, and MIT licensed. It is not an official Atlassian product. Atlassian runs its own hosted server, covered in the alternatives section below. It is a common choice for a self-hosted Atlassian MCP that also covers on-premise Jira and Confluence.
How to use Atlassian MCP Server (Jira & Confluence)?
Prerequisites
- uv installed, so that
uvx can run the package (Docker and pip also work).
- Your Jira and/or Confluence URL.
- Credentials. On Cloud, create an API token at https://id.atlassian.com/manage-profile/security/api-tokens and use it with your account email. On Server/Data Center, create a Personal Access Token from your profile and set
JIRA_PERSONAL_TOKEN / CONFLUENCE_PERSONAL_TOKEN in place of the username and API token. mTLS client certificates and OAuth 2.0 are also supported.
You can configure Jira only, Confluence only, or both. Check that the server starts with:
uvx mcp-atlassian --help
Claude Desktop
Edit claude_desktop_config.json (macOS: ~/Library/Application Support/Claude/, Windows: %APPDATA%\Claude\, Linux: ~/.config/Claude/):
{
"mcpServers": {
"mcp-atlassian": {
"command": "uvx",
"args": ["mcp-atlassian"],
"env": {
"JIRA_URL": "https://your-company.atlassian.net",
"JIRA_USERNAME": "your.email@company.com",
"JIRA_API_TOKEN": "your_api_token",
"CONFLUENCE_URL": "https://your-company.atlassian.net/wiki",
"CONFLUENCE_USERNAME": "your.email@company.com",
"CONFLUENCE_API_TOKEN": "your_api_token"
}
}
}
}
For Server/Data Center, use your own URLs (for example https://jira.your-company.com) and replace the username and API token with JIRA_PERSONAL_TOKEN / CONFLUENCE_PERSONAL_TOKEN.
Claude Code
Register the same uvx command with the CLI:
claude mcp add mcp-atlassian \
-e JIRA_URL=https://your-company.atlassian.net \
-e JIRA_USERNAME=your.email@company.com \
-e JIRA_API_TOKEN=your_api_token \
-e CONFLUENCE_URL=https://your-company.atlassian.net/wiki \
-e CONFLUENCE_USERNAME=your.email@company.com \
-e CONFLUENCE_API_TOKEN=your_api_token \
-- uvx mcp-atlassian
Cursor
Open Settings → MCP → + Add new global MCP server and paste the same JSON as for Claude Desktop.
VS Code
The project lists VS Code (GitHub Copilot) as supported over stdio. Add the uvx command as a stdio server in .vscode/mcp.json:
{
"servers": {
"mcp-atlassian": {
"type": "stdio",
"command": "uvx",
"args": ["mcp-atlassian"],
"env": {
"JIRA_URL": "https://your-company.atlassian.net",
"JIRA_USERNAME": "your.email@company.com",
"JIRA_API_TOKEN": "your_api_token"
}
}
}
}
Docker and HTTP
A Docker image is published as ghcr.io/sooperset/mcp-atlassian:latest. For shared or remote setups, run the server as an HTTP service. streamable-http serves /mcp and sse serves /sse:
uvx mcp-atlassian --transport streamable-http --port 9000 -vv
Key features of Atlassian MCP Server (Jira & Confluence)
- Works with Jira and Confluence on Cloud and on Server/Data Center (Jira 8.14+, Confluence 6.0+).
- Close to 100 tools (98 in the current docs), grouped into toolsets you can enable selectively.
- JQL search for Jira issues and CQL search for Confluence content.
- You write Markdown, and the server converts it to ADF on Cloud or wiki markup on Server/Data Center.
- Four auth methods: API token, Personal Access Token, mTLS client certificates, and OAuth 2.0.
READ_ONLY_MODE=true, ENABLED_TOOLS, TOOLSETS, JIRA_PROJECTS_FILTER and CONFLUENCE_SPACES_FILTER limit what the agent can see and change.
- stdio, SSE and streamable-http transports, including stateless mode and multi-user authentication.
- Proxy support, custom headers, and the OS trust store for corporate networks.
Atlassian MCP Server (Jira & Confluence) tools
Tools are prefixed jira_ or confluence_. Six core toolsets make up TOOLSETS=default: jira_issues, jira_fields, jira_comments, jira_transitions, confluence_pages and confluence_comments.
- Jira issues and search:
jira_search (JQL), jira_get_issue, jira_create_issue, jira_batch_create_issues, jira_update_issue, jira_assign_issue, and jira_transition_issue for status changes.
- Jira agile and projects: boards and sprints (
jira_get_agile_boards, jira_get_sprint_issues, jira_create_sprint, jira_add_issues_to_sprint), versions, components, and issue links such as jira_link_to_epic.
- Jira extras: worklogs, watchers, attachments, Jira Service Management queues and requests, SLA and issue dates, and development info.
- Confluence pages:
confluence_search (CQL), confluence_get_page, confluence_create_page, confluence_update_page, confluence_update_page_section, confluence_move_page, page history and diffs.
- Confluence collaboration: comments, inline comments, labels, attachments, templates (
confluence_create_page_from_template), and permission checks.
A few tools, such as confluence_get_page_views and jira_batch_get_changelogs, are Cloud-only.
Use cases of Atlassian MCP Server (Jira & Confluence)
- Triage your own backlog: "Find issues assigned to me in PROJ project."
- File bugs from a chat or a log: "Create a bug ticket for the login issue."
- Move work forward without opening Jira: "Update the status of PROJ-123 to Done."
- Use Confluence as a knowledge base: "Search Confluence for onboarding docs."
- Write documentation: ask the agent to draft release notes from a JQL query and publish them as a Confluence page.
Atlassian MCP Server (Jira & Confluence) vs alternatives
- Official Atlassian MCP server (Rovo): Atlassian hosts its own remote MCP server at
https://mcp.atlassian.com/v2/mcp. You sign in with OAuth 2.1 or an API token, and it covers Jira, Confluence, Jira Service Management, Bitbucket, Compass and Loom. It works only with Atlassian Cloud. Pick it if you want a vendor-supported hosted option with nothing to install. Pick mcp-atlassian if you need Server/Data Center, a self-hosted deployment, or fine-grained tool filtering.
- Atlassian Data Center MCP: a community server listed on mcp.so that targets Data Center editions of Bitbucket, Confluence and Jira.
Troubleshooting Atlassian MCP Server (Jira & Confluence)
- 401/403 on Cloud: use an API token, not your account password. Make sure
JIRA_USERNAME / CONFLUENCE_USERNAME is your email address and that the token has not expired.
- Auth failures on Server/Data Center: check that the Personal Access Token is valid. For older Confluence servers, try basic auth with
CONFLUENCE_USERNAME and CONFLUENCE_API_TOKEN, where the token is your password.
- SSL errors with self-signed certificates: enterprise CAs in the OS trust store are trusted automatically. For other self-signed certificates, set
JIRA_SSL_VERIFY=false / CONFLUENCE_SSL_VERIFY=false.
- Client shows 0 tools (for example GitHub Copilot): upgrade to
mcp-atlassian 0.16.0 or later, or run uvx mcp-atlassian@latest.
- Need more detail: set
MCP_VERBOSE=true or MCP_VERY_VERBOSE=true. In Claude Desktop on macOS, follow ~/Library/Logs/Claude/mcp*.log.
Comments