Exa Websets MCP Server

/** * Workflow guide for enriching webset data */ export async function enrichmentWorkflow(websetId: string): Promise<string> { if (!websetId) { return `Please provide a webset ID to enrich. Use \`list_websets()\` to see available websets.`; } return `# 🔮 Enrichment Workflow: ${websetId} ## 🎨 Transform Your Raw Data into Insights Enrichment adds AI-powered analysis to your webset items, turning raw web data into structured, actionable insights. ### Available Enrichment Types #### 1. 📝 Summarization Create concise summaries of each item: \`\`\` enrich_webset("${websetId}", "summarize", { style: "bullets", // "bullets" | "paragraph" | "tldr" maxLength: 200, // Characters per summary focusOn: "key_points" // What to emphasize }) \`\`\` **Use cases:** • Quick overview of long articles • Executive summaries for reports • Content preview for curation #### 2. 🏷️ Categorization Automatically categorize items: \`\`\` enrich_webset("${websetId}", "categorize", { categories: [ "Technology", "Business", "Research", "Tutorial", "News", "Other" ], multiLabel: true // Allow multiple categories per item }) \`\`\` **Use cases:** • Organize search results • Filter by category later • Understand content distribution #### 3. 🔍 Entity Extraction Extract people, companies, locations, and more: \`\`\` enrich_webset("${websetId}", "entities", { types: ["person", "organization", "location", "product"], includeMetadata: true // Add confidence scores }) \`\`\` **Use cases:** • Build relationship graphs • Track company mentions • Geographic analysis #### 4. 💭 Sentiment Analysis Analyze tone and sentiment: \`\`\` enrich_webset("${websetId}", "sentiment", { aspects: ["overall", "product", "price", "service"], includeEmotions: true // Joy, anger, fear, etc. }) \`\`\` **Use cases:** • Brand monitoring • Review analysis • Market sentiment #### 5. 🏷️ Tag Generation Generate relevant tags: \`\`\` enrich_webset("${websetId}", "tags", { maxTags: 10, style: "specific" // "broad" | "specific" | "technical" }) \`\`\` **Use cases:** • Improve searchability • Content organization • Trend identification #### 6. ❓ Q&A Extraction Extract questions and answers: \`\`\` enrich_webset("${websetId}", "qa", { extractQuestions: true, extractAnswers: true, pairMatching: true }) \`\`\` **Use cases:** • FAQ generation • Knowledge base building • Customer insight mining ### Enrichment Strategies #### Strategy 1: Layer Multiple Enrichments \`\`\`javascript // First, categorize await enrich_webset("${websetId}", "categorize", { categories: ["Technical", "Business", "Marketing"] }); // Then summarize based on category await enrich_webset("${websetId}", "summarize", { style: "bullets", customPrompts: { "Technical": "Focus on technical specifications", "Business": "Emphasize business impact", "Marketing": "Highlight market positioning" } }); \`\`\` #### Strategy 2: Conditional Enrichment \`\`\`javascript // Check content type first const items = await get_webset_items("${websetId}", 10); const hasReviews = items.some(item => item.text.includes("review") || item.text.includes("rating") ); if (hasReviews) { // Apply sentiment analysis for reviews await enrich_webset("${websetId}", "sentiment", { aspects: ["product", "service", "value"] }); } \`\`\` #### Strategy 3: Progressive Enhancement \`\`\`javascript // Start with fast enrichments await enrich_webset("${websetId}", "tags", { maxTags: 5 }); // Add detailed analysis for top items const items = await search_items("${websetId}", "", { sortBy: "relevance", limit: 50 }); await enrich_webset("${websetId}", "summarize", { itemIds: items.map(i => i.id), // Only top 50 style: "detailed" }); \`\`\` ### Monitoring Enrichment Progress Check enrichment status: \`\`\` get_webset("${websetId}") // Look for enrichments.{type}.status \`\`\` Set up webhook for completion: \`\`\` register_webhook("https://your-app.com/webhook", ["webset.enrichment.completed"], { websetId: "${websetId}" } ) \`\`\` ### Working with Enriched Data #### Access Enrichments: \`\`\`javascript const items = await get_webset_items("${websetId}", 20); items.forEach(item => { console.log("Original:", item.title); console.log("Summary:", item.enrichments?.summary); console.log("Category:", item.enrichments?.category); console.log("Sentiment:", item.enrichments?.sentiment); console.log("Entities:", item.enrichments?.entities); }); \`\`\` #### Filter by Enrichment: \`\`\` search_items("${websetId}", "", { filters: { "enrichments.category": "Technology", "enrichments.sentiment.overall": "positive" } }) \`\`\` #### Export with Enrichments: \`\`\` export_webset("${websetId}", "csv", { includeEnrichments: true, enrichmentFields: ["summary", "category", "sentiment.overall"] }) \`\`\` ### Cost Optimization Tips 1. **Sample First**: Test enrichments on a subset \`\`\` const sampleItems = await get_webset_items("${websetId}", 10); // Analyze sample before full enrichment \`\`\` 2. **Selective Enrichment**: Only enrich what you need \`\`\` enrich_webset("${websetId}", "summarize", { filter: { minTextLength: 500 } // Skip short items }) \`\`\` 3. **Reuse Enrichments**: Enrichments are cached \`\`\` // Second export uses cached enrichments export_webset("${websetId}", "json") \`\`\` ### Next Steps 1. Choose enrichments based on your use case 2. Start with a small test batch 3. Monitor progress via webhooks 4. Export enriched data for analysis Ready to enrich? Start with: \`\`\` enrich_webset("${websetId}", "summarize", { style: "bullets" }) \`\`\``; }