Pangolinfo AI SERP Skill

Google AI Mode search and standard SERP extraction with AI Overview and citations via Pangolin APIs.

When to Use This Skill

Triggers: Google SERP scraping, AI Overview extraction, AI Mode search, Google citation extraction, 抓Google SERP, 谷歌AI概览, 引用来源提取

Do not use this skill for: Amazon product searches, Google Trends, Google Maps, or non-Google search engines.

Prerequisites

• Python 3.8+ (standard library only -- no pip install needed)
• Pangolin account at pangolinfo.com

Environment Variables

Set one of:

API key resolution order: PANGOLIN_API_KEY env var > cached ~/.pangolin_api_key (if previously cached) > fresh login.

macOS SSL Fix

If you see error code SSL_CERT, run:

/Applications/Python\ 3.x/Install\ Certificates.command

Or: pip3 install certifi && export SSL_CERT_FILE=$(python3 -c "import certifi; print(certifi.where())")

Script Execution

The main script is scripts/pangolin.py relative to this skill directory.

python3 scripts/pangolin.py --q "your query"

Intent-to-Command Mapping

AI Mode Search (default)

AI-generated answers with references. Default mode, costs 2 credits.

python3 scripts/pangolin.py --q "what is quantum computing"

Standard SERP

Traditional Google search results + optional AI Overview. Costs 2 credits.

python3 scripts/pangolin.py --q "best databases 2025" --mode serp

SERP Plus (cheaper)

Same as SERP but costs only 1 credit. Uses the googleSearchPlus parser.

python3 scripts/pangolin.py --q "best databases 2025" --mode serp-plus

SERP with Screenshot

python3 scripts/pangolin.py --q "best databases 2025" --mode serp --screenshot

SERP with Region

python3 scripts/pangolin.py --q "best databases 2025" --mode serp --region us

Multi-Turn Dialogue (AI Mode only)

python3 scripts/pangolin.py --q "kubernetes" --follow-up "how to deploy" --follow-up "monitoring tools"

Auth Check (no credits consumed)

python3 scripts/pangolin.py --auth-only

All CLI Options

Supported Regions

Cost

Credits are only consumed on successful requests (API code 0). Auth checks do not consume credits.

Output Format

JSON to stdout on success, structured error JSON to stderr on failure.

Success Example (AI Mode)

{
  "success": true,
  "task_id": "1768988520324-766a695d93b57aad",
  "results_num": 1,
  "ai_overview_count": 1,
  "ai_overview": [
    {
      "content": ["Quantum computing uses quantum bits (qubits)..."],
      "references": [
        {
          "title": "Quantum Computing - Wikipedia",
          "url": "https://en.wikipedia.org/wiki/Quantum_computing",
          "domain": "Wikipedia"
        }
      ]
    }
  ]
}

Error Example (stderr)

{
  "success": false,
  "error": {
    "code": "API_ERROR",
    "message": "The search API returned an error.",
    "api_code": 2001,
    "hint": "Insufficient credits. Top up at pangolinfo.com."
  }
}

Response Presentation

1. Use natural language -- never dump raw JSON.
2. Match the user's language.
3. Summarize AI overview first, then list organic results with URLs.
4. On error, explain the issue using the hint field.

Exit Codes

Error Reference

Script Error Codes

Pangolin API Error Codes

First-Time Setup

See references/setup-guide.md for detailed first-time setup instructions.

Quick start:

export PANGOLIN_API_KEY="your-api-key"
python3 scripts/pangolin.py --auth-only

Important Notes for AI Agents

1. Run --auth-only first if unsure about credentials.
2. Default to ai-mode unless user explicitly asks for standard search.
3. Use serp-plus when user wants cheaper SERP results and doesn't need the premium parser.
4. Never expose raw JSON to the user.
5. Respond in the user's language.
6. Keep follow-ups to 5 or fewer.
7. Do not log API keys, passwords, or cookies.
8. Mention credit cost when running multiple searches.
9. --screenshot is optional -- only when user wants visual results.
10. --follow-up is ai-mode only.
11. --region is SERP/SERP-Plus only.

Output Schema

See references/output-schema.md for the complete JSON output schema.