This is not about browsing Jira; it is about closing the gap between the ticket and the change. Pull the issue, or every Bug in the project, and the original description sits in the same session as the code, so "fix PROJECT-123" has the requirement in front of it rather than someone's retyped summary. Every documented tool is a read — nothing here moves a ticket or writes a comment.
A Jira reader for Cursor and other MCP clients. It fetches one issue by key, the issues assigned to you, issues filtered by type, and what changed in a project recently — so the ticket text lands where the code is.
- One issue's full detail from its key, so the description and acceptance criteria are in the session
- The issues assigned to you within a project, capped to a result count you choose
- Issues filtered to one type — Bug, Story, Epic — across a project
- Every project the account can reach, and the issue types available in them
- What changed in a project's tickets over a window you set, such as the last 7 days
Node.js 20.17.0 or higher, a Jira account with API access, and an API token generated from Atlassian. Put JIRA_BASE_URL, JIRA_USERNAME and JIRA_API_TOKEN in .env, run npm run build, then start it. It runs as an HTTP server — clients connect to http://localhost:3000/sse by default, and HTTP_PORT moves it — with a CLI mode available as well.
One command plus a key — npx -y jira-mcp, then supply credentials