parallel-data-enrichment（并行数据丰富）

Parallel Data Enrichment

Enrich company or entity data into structured JSON using the Task API. Use core (1-5 min, $0.025) or base (15-100s, $0.01) for structured output.

CLI alternative (recommended): Install parallel-cli for official skill: npx skills add parallel-web/parallel-agent-skills --skill parallel-data-enrichment

Setup

API_KEY=$(grep "^PARALLEL_API_KEY=" .env.local | cut -d= -f2)

Create Enrichment Task

curl -s -X POST "https://api.parallel.ai/v1beta/tasks/runs" \
  -H "x-api-key: $API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "input": "OpenAI",
    "processor": "core",
    "output_schema": {
      "type": "object",
      "properties": {
        "name": {"type": "string"},
        "founded_year": {"type": "integer"},
        "headquarters": {"type": "string"},
        "employee_count": {"type": "integer"},
        "key_products": {"type": "array", "items": {"type": "string"}}
      }
    }
  }'

Response: {"id": "task_abc123", "status": "queued"}

Check Result

curl -s "https://api.parallel.ai/v1beta/tasks/runs/task_abc123" \
  -H "x-api-key: $API_KEY"

{
  "id": "task_abc123",
  "status": "completed",
  "result": {
    "content": {
      "name": "OpenAI",
      "founded_year": 2015,
      "headquarters": "San Francisco, CA",
      "employee_count": 770,
      "key_products": ["ChatGPT", "GPT-4", "DALL-E", "Whisper"]
    },
    "basis": {
      "citations": [{"url": "...", "excerpt": "..."}]
    }
  }
}

Processors

Tips for output_schema

• Keep schemas simple — fewer fields = more reliable
• Use "type": "string" broadly; avoid strict enums
• Omit optional fields from the schema

When to Use

• Company or person data enrichment
• Structured data extraction with typed schemas
• Lead qualification, CRM enrichment, research

For narrative reports, use parallel-deep-research. For web search, use parallel-web-search.