Skip to main content

Technical Documentation

Deep dive into the technical aspects of MCP JetBrains Code Inspections.

Technical Topics​

Architecture​

Comprehensive overview of the system architecture:

  • MCP server implementation
  • IDE integration mechanisms
  • Data flow and processing
  • Component interactions

Shared Inspection Engine​

Understanding JetBrains' unified inspection system:

  • How inspections work across IDEs
  • Language-agnostic architecture
  • Inspection execution pipeline
  • Result processing

IDE Inspection Equivalence​

Cross-IDE compatibility details:

  • Inspection parity between IDEs
  • Language-specific considerations
  • Profile compatibility
  • Result normalization

Inspection Consistency​

Ensuring consistent results:

  • Deterministic analysis
  • Configuration management
  • Version compatibility
  • Result stability

Key Technical Concepts​

MCP Protocol Integration​

The server provides a single tool with one required parameter:

// Server implements MCP tool interface
server.setRequestHandler(ListToolsRequestSchema, async () => ({
    tools: [
        {
            name: 'get_jetbrains_code_inspections',
            description: 'Run JetBrains IDE code inspections on files or directories',
            inputSchema: {
                type: 'object',
                properties: {
                    path: {
                        type: 'string',
                        description: 'File or directory path to inspect',
                    },
                },
                required: ['path'],
            },
        },
    ],
}));

Tool Configuration: All configuration is done via environment variables, not tool parameters:

  • FORCE_INSPECT_PATH: Force specific IDE inspect tool path
  • FORCE_PROJECT_ROOT: Force project root directory
  • FORCE_PROFILE_PATH: Force inspection profile path
  • INSPECTION_TIMEOUT: Timeout in milliseconds
  • EXCLUDE_INSPECTIONS: Comma-separated inspection codes to exclude
  • ONLY_INSPECTIONS: Comma-separated inspection codes to include only
  • RESPONSE_FORMAT: Output format ('markdown' or 'json')
  • DEBUG: Enable debug logging

IDE Detection Algorithm​

// Automatic IDE selection based on availability
const ideHierarchy = ['IntelliJ IDEA', 'WebStorm', 'PyCharm', 'PhpStorm', 'GoLand'];

Diagnostic Mapping​

// JetBrains severity to LSP diagnostic severity
const severityMap = {
    ERROR: DiagnosticSeverity.Error,
    WARNING: DiagnosticSeverity.Warning,
    'WEAK WARNING': DiagnosticSeverity.Information,
    INFO: DiagnosticSeverity.Hint,
};

Implementation Details​

Process Management​

  • Subprocess spawning for IDE tools
  • Timeout handling
  • Error recovery
  • Resource cleanup

Data Processing​

  • JSON parsing and validation
  • Path resolution
  • URI handling
  • Result transformation

Performance Considerations​

  • Lazy IDE detection
  • Result caching potential
  • Parallel processing capabilities
  • Memory management

Advanced Topics​

Custom Tool Integration​

Extending the server with additional tools

Multi-Project Support​

Handling multiple projects simultaneously

Remote Inspection​

Running inspections on remote codebases

API Extensions​

Extending the MCP protocol for custom needs

Contributing​

Technical contributions welcome:

  • Performance improvements
  • Additional IDE support
  • Enhanced error handling
  • Protocol extensions

References​