generate_context
Create context files in various formats from codebases to help AI assistants understand project structure and dependencies.
Instructions
Generate context files in specified format for the codebase
Input Schema
TableJSON Schema
| Name | Required | Description | Default |
|---|---|---|---|
| path | Yes | Root directory path | |
| format | Yes | Context file format to generate | |
| output_path | No | Output path for generated files (optional) | |
| analysis_result | No | Previous analysis result to use (optional) | |
| options | No | Format-specific options |
Implementation Reference
- src/tools/generate-context.ts:21-112 (handler)Main handler function that executes the generate_context tool logic: analyzes codebase if needed, optionally enriches with KG, generates format-specific context content, writes to file if specified, and returns success response.
export async function generateContext( args: GenerateContextArgs ): Promise<{ content: Array<{ type: string; text: string }> }> { const { path, format, output_path, analysis_result, options = {} } = args; // Get or perform analysis let analysis = analysis_result; if (!analysis) { const analyzeResult = await analyzeCodebase({ path }); analysis = JSON.parse(analyzeResult.content[0].text); } // Enrich with knowledge graph if enabled (default: true) let enriched: EnrichedContext | null = null; if (options.enable_kg !== false) { enriched = await enrichContext({ path, analysis_result: analysis, enrichment_level: options.enrichment_level || 'standard', include_yago: options.include_yago !== false, include_schema: options.include_schema !== false, max_entities: options.max_entities || 10, }); } // Generate context based on format let content: string; let filename: string; switch (format) { case 'cursorrules': content = generateCursorRules(analysis, enriched); filename = '.cursorrules'; break; case 'cursor_dir': content = generateCursorDir(analysis, enriched); filename = '.cursor/context.md'; break; case 'spec_md': content = generateSpecMd(analysis, enriched); filename = 'SPEC.md'; break; case 'agents_md': content = generateAgentsMd(analysis, enriched); filename = 'AGENTS.md'; break; case 'custom': content = generateCustom(analysis, enriched, options); filename = options.filename || 'context.md'; break; default: throw new Error(`Unknown format: ${format}`); } // Write to file if output_path specified if (output_path) { const fullPath = join(output_path, filename); const dir = dirname(fullPath); if (!existsSync(dir)) { await mkdir(dir, { recursive: true }); } await writeFile(fullPath, content, 'utf-8'); } return { content: [ { type: 'text', text: JSON.stringify( { message: 'Context generated successfully', format, path, output_file: output_path ? join(output_path, filename) : null, content_preview: content.slice(0, 500) + '...', enriched: !!enriched, kg_stats: enriched?.confidence_stats, }, null, 2 ), }, ], }; } - src/tools/generate-context.ts:13-19 (schema)TypeScript interface defining the input arguments for the generateContext handler.
interface GenerateContextArgs { path: string; format: 'cursorrules' | 'cursor_dir' | 'spec_md' | 'agents_md' | 'custom'; output_path?: string; analysis_result?: any; options?: Record<string, any>; } - src/tools/index.ts:43-44 (registration)Tool dispatch/registration in the switch statement handling CallToolRequestSchema, calling generateContext for "generate_context".
case "generate_context": return await generateContext(args as any); - src/index.ts:140-169 (registration)Tool registration in ListToolsRequestSchema response, defining name, description, and inputSchema for MCP protocol compliance.
name: "generate_context", description: "Generate context files in specified format for the codebase", inputSchema: { type: "object", properties: { path: { type: "string", description: "Root directory path", }, format: { type: "string", enum: ["cursorrules", "cursor_dir", "spec_md", "agents_md", "custom"], description: "Context file format to generate", }, output_path: { type: "string", description: "Output path for generated files (optional)", }, analysis_result: { type: "object", description: "Previous analysis result to use (optional)", }, options: { type: "object", description: "Format-specific options", }, }, required: ["path", "format"], }, }, - Helper function to generate content in '.cursorrules' format, one of several format-specific generators used by the handler.
function generateCursorRules(analysis: any, enriched: EnrichedContext | null): string { const lines: string[] = []; lines.push('# Project Rules for AI Assistants'); lines.push(''); // Project info if (analysis.package?.name) { lines.push(`## Project: ${analysis.package.name}`); if (analysis.package.description) { lines.push(analysis.package.description); } lines.push(''); } // Schema.org annotation if (enriched?.schema_annotation) { lines.push('## Semantic Type'); lines.push(`This is a **${enriched.schema_annotation['@type']}**`); if (enriched.schema_annotation.description) { lines.push(enriched.schema_annotation.description); } lines.push(''); } // Tech stack lines.push('## Tech Stack'); if (analysis.languages) { lines.push('**Languages:**', Object.keys(analysis.languages).join(', ')); } if (analysis.frameworks && analysis.frameworks.length > 0) { lines.push('**Frameworks:**', analysis.frameworks.join(', ')); } lines.push(''); // Knowledge graph enrichment if (enriched?.yago_entities && Object.keys(enriched.yago_entities).length > 0) { lines.push('## Knowledge Graph Context'); for (const [name, entities] of Object.entries(enriched.yago_entities)) { if (entities.length > 0) { const entity = entities[0]; lines.push(`- **${name}**: ${entity.description || entity.label}`); } } lines.push(''); } // Coding conventions lines.push('## Coding Conventions'); lines.push('- Follow existing patterns in the codebase'); lines.push('- Use TypeScript strict mode'); lines.push('- Write clear, self-documenting code'); lines.push('- Add JSDoc comments for public APIs'); lines.push(''); return lines.join('\n'); }