MCP server for SonarQube integration - provides tools to query project analysis, quality gates, metrics, issues, and run code scans
17 tools with complete schemas and descriptions, but quality is uneven. Most descriptions are adequate (50-150 chars) and follow verb_noun naming. However, parameter descriptions lack depth, many omit constraints, ranges, and format guidance. Output schemas are undocumented. Error handling is minimal. sonar_call_multiple violates single-responsibility principle. No tool annotations (readOnlyHint/destructiveHint). Baseline: 549 production tools average 194 chars per description and 72 chars per param; this server averages ~100 chars per description and ~40 chars per param.
Check if a project has been analyzed. Examples: sonar_analysis_status({ projectKey: "my_proj" })
Execute multiple tool calls in sequence. Useful for complex workflows. Examples: sonar_call_multiple({ calls: [{ name: "sonar_ping", args: {} }, { name: "sonar_summary", args: { projectKey: "my_proj" } }] })
Get detailed information about a security hotspot. Requires User token (squ_ prefix). Examples: sonar_hotspot_details({ hotspotKey: "AXpg2HiGIW-bsFB_5L7x" })
Search security hotspots. Requires User token (squ_ prefix). Examples: sonar_hotspots_search({ projectKey: "my_proj" })
Get detailed information about a specific issue including code context. Examples: sonar_issue_details({ issueKey: "AVF1R6dJV-G6OnNq2HCs" })
Search issues with filtering and facets. Examples: sonar_issues_search({ projectKey: "my_proj" }), sonar_issues_search({ projectKey: "my_proj", types: "BUG,VULNERABILITY" })
sonar_call_multiple violates single-responsibility principle. A tool that executes arbitrary sequences of other tools is a meta-tool that bypasses the agent's planning layer. This prevents proper error handling, retry logic, and observability per call.
Parameter descriptions lack actionable constraints. E.g., 'limit' params say 'default 50, max 500' but do not specify minimum or explain why the max exists. 'branch' and 'pullRequest' are optional but never explain when to use each or if they are mutually exclusive.
No output schemas documented. Tools return JSON but LLMs cannot plan downstream calls without knowing field names and types. E.g., sonar_issues_search returns issues but the response structure is invisible to the agent.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 53 | <=2025-11-25 | v2 |
List all quality gates. Examples: sonar_list_quality_gates({})
Get metrics: bugs, smells, coverage, ratings, ncloc, dup. Examples: sonar_measures({ projectKey: "my_proj" }), sonar_measures({ projectKey: "my_proj", metricKeys: "coverage,bugs" })
Ping server health — returns pong + status. Examples: sonar_ping({})
Get detailed information about a project. Examples: sonar_project_details({ projectKey: "my_proj" })
Create a new project in SonarQube. Requires admin permissions. Examples: sonar_projects_create({ projectKey: "new_project" }), sonar_projects_create({ projectKey: "new_project", name: "My Project" })
Get QG status with failing conditions. Examples: sonar_quality_gate({ projectKey: "my_proj" })
Escape hatch — call any GET endpoint. Requires appropriate token permissions for the endpoint. Examples: sonar_raw({ path: "/api/system/health" })
Run SonarQube scanner on a local project directory. Requires Docker or sonar-scanner CLI. Examples: sonar_scan({ projectKey: "my_proj", directory: "." }), sonar_scan({ projectKey: "my_proj", directory: ".", language: "python" })
Browse available metric definitions. Examples: sonar_search_metrics({ query: "coverage" })
Search/find SonarQube project keys. Examples: sonar_search_projects({ query: "my" }), sonar_search_projects({ limit: 10 })
Get aggregated project health: QG, metrics, issues, branches. Examples: sonar_summary({ projectKey: "my_proj" })
No tool annotations (readOnlyHint, destructiveHint, idempotentHint). Agents cannot distinguish safe reads from destructive writes. sonar_projects_create and sonar_scan are marked WRITE but lack destructiveHint annotation in the schema.
sonar_raw is a generic escape hatch that accepts any /api/ path. This bypasses schema validation and error handling, making it impossible for agents to reason about safety or expected output. Prefer explicit tools for each API endpoint.