analyze-cursor
Analyzes code at a specific cursor position to provide context-aware insights and identify relevant patterns or issues in the surrounding code structure.
Input Schema
TableJSON Schema
| Name | Required | Description | Default |
|---|---|---|---|
| filePath | Yes | Path to the file | |
| fileContent | Yes | Content of the file | |
| line | Yes | Cursor line position (1-based) | |
| column | Yes | Cursor column position (1-based) | |
| context | Yes | Context around cursor |
Implementation Reference
- Handler function that processes the cursor position context, identifies the code entity using helper functions, analyzes it, and returns a JSON-formatted response.async (args: { filePath: string; fileContent: string; line: number; column: number; context: { before: string[]; target: string; after: string[]; } }) => { const { filePath, fileContent, line, column, context } = args; // Determine language from file extension const extension = path.extname(filePath).slice(1); // Find the code entity at the cursor position const entity = identifyCodeEntity(fileContent, line, column, extension); // Analyze the entity const analysis = analyzeCodeEntity(entity, extension); return { content: [{ type: "text", text: JSON.stringify({ file: path.basename(filePath), position: { line, column }, entity: entity.type, name: entity.name, analysis }, null, 2) }] }; }
- Zod schema defining the input parameters for the analyze-cursor tool, including file info, position, and local context.{ filePath: z.string().describe("Path to the file"), fileContent: z.string().describe("Content of the file"), line: z.number().describe("Cursor line position (1-based)"), column: z.number().describe("Cursor column position (1-based)"), context: z.object({ before: z.array(z.string()).describe("Lines before cursor"), target: z.string().describe("Line at cursor"), after: z.array(z.string()).describe("Lines after cursor") }).describe("Context around cursor") },
- src/features/basic-analysis/ide-analyzer.ts:7-55 (registration)Registration of the analyze-cursor tool on the MCP server using server.tool() with name, input schema, and handler function.server.tool( "analyze-cursor", { filePath: z.string().describe("Path to the file"), fileContent: z.string().describe("Content of the file"), line: z.number().describe("Cursor line position (1-based)"), column: z.number().describe("Cursor column position (1-based)"), context: z.object({ before: z.array(z.string()).describe("Lines before cursor"), target: z.string().describe("Line at cursor"), after: z.array(z.string()).describe("Lines after cursor") }).describe("Context around cursor") }, async (args: { filePath: string; fileContent: string; line: number; column: number; context: { before: string[]; target: string; after: string[]; } }) => { const { filePath, fileContent, line, column, context } = args; // Determine language from file extension const extension = path.extname(filePath).slice(1); // Find the code entity at the cursor position const entity = identifyCodeEntity(fileContent, line, column, extension); // Analyze the entity const analysis = analyzeCodeEntity(entity, extension); return { content: [{ type: "text", text: JSON.stringify({ file: path.basename(filePath), position: { line, column }, entity: entity.type, name: entity.name, analysis }, null, 2) }] }; } );
- Placeholder helper to identify the type and details of the code entity at the cursor position.function identifyCodeEntity( content: string, line: number, column: number, language: string ): { type: string; name: string; startLine: number; endLine: number; complexity: number; } { // This is a placeholder implementation // In a real implementation, you would use a language parser to identify // the function, class, or other entity at the cursor position return { type: "function", name: "exampleFunction", startLine: line - 2, endLine: line + 5, complexity: 3 }; }
- Placeholder helper to analyze the identified code entity and provide metrics and recommendations.function analyzeCodeEntity( entity: { type: string; name: string; startLine: number; endLine: number; complexity: number }, language: string ): { complexity: number; linesOfCode: number; recommendations: string[]; } { // This is a placeholder implementation // In a real implementation, you would perform deeper analysis on the entity return { complexity: entity.complexity, linesOfCode: entity.endLine - entity.startLine + 1, recommendations: [ "Consider breaking this function into smaller parts", "Add more descriptive variable names" ] }; }