mcp-rtfm

Instructions

Implementation Reference

• src/index.ts:741-866 (handler)

  Main execution logic for the 'analyze_project_with_metadata' tool. Initializes .handoff_docs directory, creates default MD files if missing, analyzes content with remark, categorizes, generates metadata, finds related docs, updates search index, adds YAML frontmatter, and returns project status with git info.
  case "analyze_project_with_metadata": { const { projectPath } = request.params.arguments as { projectPath: string }; const docsPath = `${projectPath}/.handoff_docs`; try { // First run the standard analyze_project workflow await fs.mkdir(docsPath, { recursive: true }); // Initialize default documentation files if they don't exist for (const doc of DEFAULT_DOCS) { const filePath = `${docsPath}/${doc}`; try { await fs.access(filePath); } catch { const title = doc.replace(".md", "") .split(/[_-]/) .map(word => word.charAt(0).toUpperCase() + word.slice(1)) .join(" "); await fs.writeFile(filePath, TEMPLATE_CONTENT.replace("{title}", title)); } } // Reset state state = { currentFile: null, completedFiles: [], inProgress: false, lastReadFile: null, lastReadContent: null, continueToNext: false, metadata: {}, contextCache: {}, templateOverrides: {} }; // Clear existing search index searchEngine.removeAll(); // Now enhance each file with metadata and context for (const doc of DEFAULT_DOCS) { const filePath = `${docsPath}/${doc}`; const content = await fs.readFile(filePath, "utf8"); // Use unified/remark to analyze content structure const analysis = await analyzeContent(content); // Use enhanced categorization const { category, tags } = categorizeContent(doc, content, analysis); // Generate metadata const metadata = { title: analysis.title || doc.replace(".md", "") .split(/[_-]/) .map(word => word.charAt(0).toUpperCase() + word.slice(1)) .join(" "), category, tags, lastUpdated: new Date().toISOString() }; // Update metadata for the file await updateMetadata(filePath, metadata); // Find and update related docs const relatedDocs = await findRelatedDocs(doc, projectPath); await updateMetadata(filePath, { relatedDocs }); // Update search index with full content and metadata updateSearchIndex(doc, content, { ...metadata, relatedDocs }); // Add structured front matter to content const enhancedContent = `--- title: ${metadata.title} category: ${metadata.category} tags: ${metadata.tags.join(', ')} lastUpdated: ${metadata.lastUpdated} relatedDocs: ${relatedDocs.join(', ')} --- ${content}`; // Update file with enhanced content await fs.writeFile(filePath, enhancedContent); } // Get project info for additional context let gitInfo = {}; try { gitInfo = { remoteUrl: execSync("git config --get remote.origin.url", { cwd: projectPath }).toString().trim(), branch: execSync("git branch --show-current", { cwd: projectPath }).toString().trim(), lastCommit: execSync("git log -1 --format=%H", { cwd: projectPath }).toString().trim() }; } catch { // Not a git repository or git not available } return { content: [ { type: "text", text: JSON.stringify({ message: "Documentation structure initialized with metadata and context", docsPath, files: DEFAULT_DOCS, metadata: state.metadata, gitInfo, contextCache: { timestamp: state.contextCache.timestamp, ttl: CACHE_TTL } }, null, 2) } ] }; } catch (error: unknown) { const errorMessage = error instanceof Error ? error.message : String(error); throw new McpError( ErrorCode.InternalError, `Error initializing documentation with metadata: ${errorMessage}` ); } }
• src/index.ts:404-417 (registration)

  Tool registration in ListToolsRequestSchema handler, including name, description, and input schema definition.
  { name: "analyze_project_with_metadata", description: "Analyze project structure, create initial documentation files, and enhance with metadata/context", inputSchema: { type: "object", properties: { projectPath: { type: "string", description: "Path to the project root directory" } }, required: ["projectPath"] } },
• src/index.ts:407-416 (schema)

  Input schema definition for the tool, specifying projectPath as required string parameter.
  inputSchema: { type: "object", properties: { projectPath: { type: "string", description: "Path to the project root directory" } }, required: ["projectPath"] }