SonarQube MCP

Sonar's MCP server connects an AI client to SonarQube Server or SonarQube Cloud so an assistant can read the same static analysis findings a developer sees in the web interface. It can search issues filtered by project, component, severity, or rule, retrieve the rule description and remediation guidance behind a finding, check the quality gate status of a project or branch, and pull measures such as coverage, duplication, and issue counts. Writing is supported where the token allows it, so an agent can change an issue's status, for example marking it accepted or a false positive with a justification comment. The server can also run Sonar analyzers over a code snippet locally, which lets an assistant check code it just generated against the team's rule set before that code reaches a commit or a pull request. Authentication is a user token plus the server URL, with an organization key on SonarQube Cloud, and the agent only reaches the projects that token can already see. For engineering teams that gate merges on Sonar, this closes the loop between the analyzer and the assistant writing the code: instead of a developer relaying failures back into a chat window, the agent reads the quality gate, pulls the offending issues with their rule context, and proposes a fix in the same session.