Skip to main content

Quick Start

Get started with MCP JetBrains Code Inspections in just a few minutes.

Basic Usage​

Once you have installed the server, you can start using it immediately.

Using with Claude Code​

  1. Ensure the server is configured in your MCP settings
  2. Restart Claude Code to load the new server
  3. The get_jetbrains_code_inspections tool will be available for code analysis

Direct Usage Example​

// Analyze a TypeScript file
const result = await get_jetbrains_code_inspections({
    path: '/path/to/your/file.ts',
});

// Analyze a Python file
const result = await get_jetbrains_code_inspections({
    path: '/path/to/your/script.py',
});

// Analyze an entire directory
const result = await get_jetbrains_code_inspections({
    path: '/path/to/your/project',
});

Understanding the Output​

The server supports two output formats: Markdown (default) and JSON. You can configure the format using the RESPONSE_FORMAT environment variable.

Output Format Configuration​

Configure the response format in your .mcp.json:

{
    "mcpServers": {
        "jetbrains-inspections": {
            "command": "node",
            "args": ["/path/to/dist/index.js"],
            "env": {
                "RESPONSE_FORMAT": "markdown"
                // or "json"
            }
        }
    }
}

Markdown Format (Default)​

When RESPONSE_FORMAT is set to "markdown" (or not set), the output is formatted as human-readable Markdown:

⚠️ **Issues found** - 67 problems detected (2 errors, 63 warnings, 2 infos).

## Code Inspection Results

**Total Issues**: 67

- 🔴 Errors: 2
- 🟡 Warnings: 63
- 🔵 Info: 2

---

### 🔴 Errors

#### 📄 `scripts/testing/example-errors-showcase.ts`

- **Line 12, Column 41**
    - Code: `Annotator`
    - Annotator: , or ) expected #loc

- **Line 298, Column 17** (length: 1)
    - Code: `Annotator`
    - Annotator: Unclosed character class #loc
    - Element: `'`

### 🟡 Warnings

#### 📄 `scripts/testing/example-errors-showcase.ts`

- **Line 11, Column 9** (length: 14)
    - Code: `JSUnusedGlobalSymbols`
    - JSUnusedGlobalSymbols: Unused function brokenFunction
    - Element: `brokenFunction`

- **Line 40, Column 4** (length: 38)
    - Code: `UnreachableCodeJS`
    - UnreachableCodeJS: Unreachable code #loc
    - Element: `console.log("This will never execute")`

### 🔵 Information

#### 📄 `scripts/testing/example-errors-showcase.ts`

- **Line 192, Column 1** (length: 5)
    - Code: `JSIgnoredPromiseFromCall`
    - JSIgnoredPromiseFromCall: Promise returned from fetch is ignored
    - Element: `fetch`

Special cases:

  • No issues found: ✅ **No issues found** - Code inspection completed successfully with no problems detected.
  • Errors: ❌ **Inspection Error**\n\n{error message}
  • Warnings: ⚠️ **Warning**: {warning message}\n\n{totalProblems} problems found

JSON Format​

When RESPONSE_FORMAT is set to "json", the output is structured JSON:

{
    "totalProblems": 67,
    "diagnostics": [
        {
            "file": "scripts/testing/example-errors-showcase.ts",
            "line": 12,
            "column": 41,
            "severity": "error",
            "code": "Annotator",
            "message": "Annotator: , or ) expected #loc"
        },
        {
            "file": "scripts/testing/example-errors-showcase.ts",
            "line": 298,
            "column": 17,
            "length": 1,
            "severity": "error",
            "code": "Annotator",
            "message": "Annotator: Unclosed character class #loc",
            "highlightedElement": "'"
        },
        {
            "file": "scripts/testing/example-errors-showcase.ts",
            "line": 279,
            "column": 1,
            "length": 55,
            "severity": "warning",
            "code": "ES6UnusedImports",
            "message": "ES6UnusedImports: <html>Unused <pre><code>...</code></pre></html>",
            "highlightedElement": "import { circularDep } from './test-inspection-errors';"
        }
    ]
}

Note: When errors occur or metadata is available, additional fields may be present:

{
    "totalProblems": 0,
    "diagnostics": [],
    "error": "Error message if inspection failed",
    "warning": "Warning message if partially completed",
    "timeout": false,
    "metadata": {
        "targetPath": "/path/to/target",
        "projectRoot": "/path/to/project",
        "ideUsed": "WebStorm",
        "ideVersion": "2024.3",
        "executionTime": 14271,
        "timestamp": "2025-08-09T05:13:30.669Z"
    }
}

Severity Levels​

  • error - Critical issues that must be fixed (compilation errors, syntax errors)
  • warning - Issues that should be addressed (unused code, potential bugs, code quality)
  • info - Information, suggestions and minor issues (typos, style suggestions)

Diagnostic Fields​

Core fields in each diagnostic object:

  • file - Path to the file containing the issue (relative or absolute)
  • line - Line number where the issue starts (1-indexed)
  • column - Column number where the issue starts (1-indexed)
  • severity - Issue severity level ("error", "warning", or "info")
  • message - Full message including inspection code and description

Optional fields that may appear:

  • length - Length of the highlighted region in characters
  • code - JetBrains inspection code (e.g., "JSUnusedLocalSymbols")
  • highlightedElement - The specific code element with the issue
  • category - JetBrains inspection category (rarely present)
  • hints - Array of strings with fix suggestions (rarely present)

Root-level fields in the response:

  • totalProblems - Total count of detected issues
  • diagnostics - Array of diagnostic objects
  • error - Error message if the inspection failed (optional)
  • warning - Warning message if inspection partially completed (optional)
  • timeout - Boolean indicating if inspection timed out (optional)
  • metadata - Additional execution metadata (optional, rarely included)

Common Use Cases​

1. Check a Single File​

await get_jetbrains_code_inspections({
    path: '/src/components/Button.tsx',
});

2. Analyze a Directory​

// This will analyze all files in the directory
await get_jetbrains_code_inspections({
    path: '/src',
});

3. Configure Output Format​

Configure the response format using environment variables in your .mcp.json:

{
    "mcpServers": {
        "jetbrains-inspections": {
            "command": "node",
            "args": ["/path/to/dist/index.js"],
            "env": {
                "RESPONSE_FORMAT": "json",
                "DEBUG": "true"
            }
        }
    }
}

Testing Your Setup​

Using the Test Script​

# Run interactive MCP test
yarn test:mcp
yarn test:mcp --path scripts/testing/example-errors-showcase.ts

This interactive script allows you to:

  • Test the MCP server functionality
  • Input file paths to inspect
  • View formatted inspection results
  • Test different configuration options

Using MCP Inspector​

npx @modelcontextprotocol/inspector dist/index.js

Then in the inspector:

  1. Connect to the server
  2. Call get_jetbrains_code_inspections with a file path
  3. Review the returned diagnostics

Troubleshooting Quick Start Issues​

If you encounter issues:

  1. Check IDE is installed: The server needs at least one JetBrains IDE
  2. Verify file paths: Use absolute file or directory paths
  3. Check timeouts: Large files may need increased timeout values
  4. Review logs: Enable debug logging for detailed information

Next Steps​