Pangolinfo AI SERP: AI Mode Output + AI Overviews

The Problem

When engineering teams need Google search results that combine AI-generated answers, traditional SERP, AI Overviews, and citations, scraping raw pages is fragile: login states, regional layouts, anti-bot behavior, and changing AI Overview markup all complicate parsing. Pangolinfo AI SERP turns this into a scriptable API workflow, so the consumer does not have to rebuild page logic for each request.

How It Works

The skill centers on scripts/pangolin.py and exposes three extraction paths: ai-mode returns AI-generated answers with references, serp returns standard Google results, and serp-plus uses a lower-cost parser. Key capabilities include:
- Query and result extraction: supports --q, --num, and --region for different locales and result counts.
- AI Overview and citations: summarizes AI Mode answers while preserving source links for verification.
- Follow-up turns: --follow-up is limited to AI Mode and is useful for narrowing the same search context.
- Stable I/O: success writes JSON to stdout, failure writes structured errors to stderr, which makes automation and agent pipelines easier to handle.

Boundaries and Notes

Use it only for Google SERP, AI Overviews, and citation extraction; do not use it for Amazon product searches, Google Trends, Maps, or non-Google engines. Configure PANGOLIN_API_KEY or account credentials before running. Different modes consume different credits, and only successful requests are billed. If SSL_CERT appears, fix the local certificate environment; if RATE_LIMIT or account errors appear, reduce request frequency or check account status.