Sonar MCP Server

Sonar MCP Server

[CI] [Release] [License] [Java 25] [MCP] [Glama score] [Listed on mcpservers.org]

A local MCP server providing read-only access to a SonarQube Community Build (26.4+) instance via its web-api. It lets AI agents (Claude Code, Cursor, VS Code Copilot, etc.) fetch a project's issue list, the files and locations where they occur, rule descriptions, source-code snippets around issues, and Security Hotspots.

Typical scenario: "find and fix Sonar issues in such-and-such project" — the LLM calls listIssues, optionally getRule and getIssueSnippets, and edits files locally.

Why this server

There is an official SonarSource MCP server, but it targets SonarQube 10+ (and SonarCloud) and assumes a cloud-style deployment. This server is built for self-hosted SonarQube Community Build 26.4+ installations that expose the classic /api/ web-api. It supports both Standard Experience and MQR mode, handling per-software-quality impacts when the instance runs in MQR mode.

It is also intentionally narrower in scope:

• Read-only by design. The server never creates, updates, or deletes anything in SonarQube — no marking issues as false-positive, no editing comments, no admin endpoints. The token's write permissions in SonarQube are irrelevant because the server never calls those endpoints.
• Curated tool set. Instead of mirroring the SonarQube API surface, the server exposes a small, focused set of tools (13 in total) chosen for a single workflow: let an AI agent read Sonar's findings and fix the code based on them. Listing components, issues and hotspots, drilling into a single finding, fetching the rule explanation, and pulling the source-code snippet around the location — and that's it. Anything outside this "diagnose -> fix the code locally" loop is deliberately left out to keep the tool list small and the agent's choices unambiguous.

In short: a focused, read-only bridge from a self-hosted SonarQube Community Build to an AI coding agent.

Quick start

1. Install JDK 25+.
2. Download sonar-mcp-server.jar from the latest release, or build it yourself: ./gradlew bootJar (see Build). A Docker image is published as well.
3. Get your SonarQube URL and user token (see Configuration).
4. Add the JAR to your client's MCP configuration (see Connecting to an AI client).

For Claude Code that is one command:

claude mcp add --scope user -e SONAR_URL=https://sonar.example.com -e SONAR_TOKEN=your_token -- sonar java -jar /path/to/sonar-mcp-server.jar

Architecture

The server only supports the stdio transport.

┌─────────────┐     stdio      ┌──────────────────┐    web-api     ┌──────────┐│  AI agent   │ <------------> │  sonar-mcp-      │ -------------> │ SonarQube││ (Claude Code│   stdin/stdout │  server (Java)   │  HTTP + Bearer  │  CB 26.4+││  Cursor...) │                │                  │  auth (token)   │          │└─────────────┘                └──────────────────┘                └──────────┘

The AI client spawns the server as a child process; communication uses the MCP protocol over stdin/stdout. The server does not open any HTTP port and accepts no incoming connections.

Tools

The server exports 13 read-only MCP tools.

Projects

Issues

Rules

Security Hotspots

All tools are read-only — no data in SonarQube is modified.

Working with branches and pull requests

Sonar analyses a branch and a pull request as two distinct, mutually exclusive scopes. The Sonar web-api accepts either branch= or pullRequest= on a single request, never both.

• branch — long-lived branches (main, develop, feature/...). Resolved as: explicit branch argument → SONAR_DEFAULT_BRANCH → none (Sonar uses the project's main branch).
• pullRequest — the Sonar PR key, usually the PR/MR number. Independent from branch analyses; PR analyses often contain the most relevant findings for in-flight work. Pull request keys never fall back to a server-level default — pass them explicitly.

Passing both branch and pullRequest to the same tool call is an error. Use listProjectBranches / listProjectPullRequests to discover available refs.

Branch scoping is load-bearing. Each Sonar branch is a separate analysis: open-issue counts on main and on a feature branch can differ a lot, because the feature branch may have new fixes (or new code) that have not been merged. If the agent silently omits branch=, it reads from main (or SONAR_DEFAULT_BRANCH) and may miss work already done on a feature branch — or worse, fix the same problem twice. When the user's local git is on feature/..., agents should call listProjectBranches first, find the matching Sonar branch, and pass it explicitly to every issue / summary / breakdown call.

As a server-side safety net, listIssues, getProjectIssuesSummary, and getProjectIssuesBreakdown attach a branchAdvisory field to their response when (1) neither branch nor pullRequest was passed AND (2) the project has other branches analysed in Sonar. The advisory carries the effective (main) branch name and the list of non-main branches sorted by most recent analysisDate. The field is absent otherwise.

Stack

• Java 25, Spring Boot 4.0.0, Spring AI MCP 2.0.0 (stdio transport)
• Jackson 3 (tools.jackson) for JSON
• Gradle 9.3.1 with version catalog (gradle/libs.versions.toml)

Build

# Point to a JDK 25+ if it's not the default:export JAVA_HOME="$HOME/.jdks/jdk-25.0.2"
./gradlew build

On Windows: .\gradlew.bat build.

Output: build/libs/sonar-mcp-server.jar

Configuration

The server needs a SonarQube URL and token; the rest is optional.

Tool groups

The MCP tools/list manifest is split into four groups, each gated by an environment flag. All groups are on by default, so the out-of-the-box manifest is unchanged. Turn a group off to shrink the manifest — useful for small-context (local) models that would otherwise be flooded with tool and output schemas at session start.

Set a flag to false (e.g. SONAR_MCP_TOOLS_HOTSPOT=false) to drop that group's tools from the manifest.

Getting the SonarQube URL

Open SonarQube in a browser and copy the address from the location bar without the path — only scheme and host.

If SonarQube is reachable only by IP, use the IP as is. If it's deployed under a subpath (e.g. /sonar), include that in the URL as well.

Getting a SonarQube token

1. Sign in to SonarQube with your account.
2. Open My Account -> Security.
3. In Generate Tokens, enter a token name and pick type User Token.
4. Click Generate — the token is shown only once. Copy it immediately.
5. Use the token value as SONAR_TOKEN.

If a token is lost, you have to regenerate it — SonarQube doesn't display existing tokens again.

The server uses HTTP Bearer auth, passing the token in the Authorization header — this is the standard SonarQube Community Build scheme for user tokens.

Running

SONAR_URL=http://sonar.example.com SONAR_TOKEN=your_token \  java -jar build/libs/sonar-mcp-server.jar

The server runs over stdio. After a successful start it opens no HTTP port and waits for MCP requests over stdin/stdout.

Logs are written to ${SONAR_MCP_DATA_DIR:-~/.sonar-mcp-server}/logs/sonar-mcp-server.log. The file rotates by date and size: 10MB, retention 30 days, total cap 512MB.

Docker

The image is published to GHCR with every release:

docker run -i --rm   -e SONAR_URL=https://sonar.example.com   -e SONAR_TOKEN=your_token   ghcr.io/igorolv/sonar-mcp-server:latest

The same command is what an MCP client should launch (-i keeps stdin open for the stdio transport). Mount a host directory at /data to keep logs between runs. If SonarQube runs on the same machine, use its host name rather than localhost, or add --network host on Linux. To build the image locally: docker build -t sonar-mcp-server .

Connecting to an AI client

{  "command": "java",  "args": ["-jar", "<absolute-path>/sonar-mcp-server.jar"],  "env": {    "SONAR_URL": "http://sonar.example.com",    "SONAR_TOKEN": "your_token"  }}

Where exactly:

For CLI clients there are also commands to view and remove the registration: claude mcp list / claude mcp remove --scope user sonar (similarly for qwen).

After adding, restart the client.

Example AI-agent prompts

Find and fix Sonar issues in project my-projectShow the top 10 rules by number of open issues in project my-projectShow Sonar issues under src/main/java/com/example/foo

Operations and security

This MCP server is meant to run locally next to the AI client. It opens no HTTP port and accepts no incoming network connections: the client starts the JAR as a child process and talks to it over stdin/stdout.

Access model

• The server acts with the rights of the SonarQube user whose token is in SONAR_TOKEN.
• All MCP tools are read-only: the server does not create, modify, or delete issues, hotspots, rules, or projects.
• Available projects and issues are determined by the user's permissions in SonarQube. If the user can't see a project in SonarQube, the server shouldn't be able to access it either.
• Treat the token as a secret. Don't commit it to the repository, shell scripts, .vscode/mcp.json, .cursor/mcp.json, or any other shared project files.

What data is sent to the AI client

The AI client receives exactly the data it requests through the MCP tools:

• the list of issues with rule, severity, type, file path, line number, message, tags, SCM author;
• issue change history;
• Sonar rule descriptions (including HTML/markdown);
• source-code snippets around issue locations (via getIssueSnippets);
• Security Hotspots and their details.

Before connecting to an external or cloud-based AI client, check your company's internal policies: source code may contain trade secrets.

Diagnostics

Environment check:

java -versionecho "$SONAR_URL"test -n "$SONAR_TOKEN" && echo "SONAR_TOKEN is set"

SonarQube web-api access check (HTTP Bearer auth with the token):

curl -H "Authorization: Bearer $SONAR_TOKEN" "$SONAR_URL/api/components/search?qualifiers=TRK&p=1&ps=1"

The expected response is a JSON list of projects. 401 Unauthorized means the token is invalid or expired; 403 means the user lacks permission for the API.

Build check:

./gradlew test./gradlew build

Integration tests against a live SonarQube:

SONAR_URL=<url> SONAR_TOKEN=<token> ./gradlew integrationTest

Integration tests require a reachable SonarQube and real data. Unit tests exclude the integration tag by default.

Known operational limitations

• HTTP timeouts and retry policy aren't separately configurable yet.
• The Sonar API uses page-based pagination (p/ps); the tools accept offset/limit, and an offset that is not a multiple of limit is rounded down to the nearest page boundary. Sonar also caps ps at 500.
• componentPathPrefix is a client-side filter: the server pages through the project's issues / hotspots and keeps those whose componentPath starts with the prefix (directory-boundary safe). The scan is capped (SONAR_MCP_PATH_FILTER_MAX_SCANNED_ISSUES, default 10000); when the cap is hit, the response sets pathPrefixTruncated=true and the caller should tighten the prefix.
• The author field on Issue is the SCM author of the line where the issue occurred (populated by Sonar when an SCM provider is configured). Sonar doesn't return separate scmAuthor/scmDate fields in issues/search; for line-level SCM use getIssueSnippets.

Project layout

├── src/main/java/ru/it_spectrum/ai/sonar/mcp/│   ├── SonarMcpServerApplication.java   — Spring Boot entry point│   ├── api/                              — stable MCP wire format: records returned by tools/services│   │   ├── Issue.java, IssuePage.java, IssueDetails.java, IssueLocation.java, IssueFlow.java, IssueImpact.java, IssueSnippets.java│   │   ├── Project.java, ProjectPage.java, ProjectOverview.java, ProjectMetrics.java, ProjectComponent.java, ProjectComponentPage.java│   │   ├── RuleDetails.java, RuleSection.java│   │   ├── Hotspot.java, HotspotDetails.java, HotspotPage.java, HotspotRule.java│   │   ├── SourceSnippet.java, SnippetLine.java│   │   ├── ChangelogEntry.java, ChangelogDiff.java│   │   ├── TextRange.java, FacetCount.java, BranchAdvisory.java│   │   ├── ProjectIssuesSummary.java, ProjectIssuesBreakdown.java, ModuleIssuesSummary.java│   │   ├── ProjectBranch.java, ProjectBranches.java, ProjectPullRequest.java, ProjectPullRequests.java│   │   ├── QualityGateStatus.java, QualityGateCondition.java│   │   └── Opaque.java│   ├── client/│   │   ├── SonarClient.java              — SonarQube web-api wrapper│   │   └── model/                        — raw DTOs of the SonarQube web-api, not exposed directly via MCP│   │       └── Sonar*.java│   ├── config/│   │   ├── SonarClientProperties.java   — url + token from env│   │   ├── SonarMcpProperties.java      — all sonar-mcp.* runtime settings│   │   ├── SonarConfig.java             — RestClient with Bearer auth│   │   ├── McpServerConfig.java         — stdio MCP customizer with immediateExecution(true)│   │   └── JsonConfig.java              — ObjectMapper for MCP JSON│   ├── service/│   │   ├── ProjectService.java│   │   ├── IssueService.java│   │   ├── RuleService.java             — with in-memory rule cache│   │   ├── SnippetService.java│   │   ├── HotspotService.java│   │   ├── PaginationHelper.java        — offset/limit -> p/ps│   │   └── SonarMappers.java            — client.model -> api mapping│   └── tools/│       ├── ProjectTools.java            — 5 MCP tools│       ├── IssueTools.java              — 5 MCP tools│       ├── RuleTools.java               — 1 MCP tool│       ├── HotspotTools.java            — 2 MCP tools│       ├── ToolDescriptions.java        — shared @McpTool / @McpToolParam description constants│       ├── SonarPrompts.java            — MCP prompts (analyzePath, fixPath, fixFile, investigateIssue, reviewPullRequest)│       ├── RefResolver.java             — branch/pullRequest resolution with default-branch fallback│       └── ToolLogger.java└── src/main/resources/    ├── application.yml                  — MCP server configuration (stdio)    └── logback-spring.xml               — logging configuration

Troubleshooting

• "Gradle requires JVM 17 or later" — set JAVA_HOME to a JDK 25+.
• Connection refused / 401 — check URL and token. Test: curl -H "Authorization: Bearer $SONAR_TOKEN" "$SONAR_URL/api/components/search?qualifiers=TRK&p=1&ps=1".
• 403 Forbidden — the token user has no rights on the project or on the web-api. Check the role in SonarQube.
• Package/module scope returns 0 issues — pass the path-style prefix as componentPathPrefix (e.g. bc-doc/src/main/java/ru/foo), not a Sonar component key. Convert Java/Kotlin package dots to slashes. The filter is directory-boundary safe, so bc-doc/src will not match bc-doc/srcExtra.
• pathPrefixTruncated=true in the response — the client-side scan hit SONAR_MCP_PATH_FILTER_MAX_SCANNED_ISSUES before reaching the end. Tighten the prefix to reduce the scan, or raise the cap if 10k is genuinely not enough for your project.