external-reference-verification

What problem does it solve?

This skill solves the critical problem of inaccurate fact verification when named authoritative sources are unreachable, blocked by bot protection like Cloudflare or JavaScript challenges, or when search engine AI summaries provide incorrect values. It eliminates the risk of silently downgrading to weaker, unvetted sources by enforcing a strict source acquisition and fallback workflow.

Core Features & Use Cases

• Ranked Fallback Source Hierarchy: Automatically tries the named authoritative source first, then independent mirrors, official mirrored PDFs, and only uses general web search to locate sources (never as the value itself).
• Per-Source Extraction Adapters: Custom parsing logic for Next.js sites (extracting NEXT_DATA JSON blobs), official PDFs (extracting text layers), and plain HTML to pull only the target data field.
• Independent Confirmation Requirement: Enforces at least one second authoritative source confirmation per data point to avoid errors from single scraped sources.
• Use Case: For example, if you need to verify an answer key from a publisher's site that is blocked by a Cloudflare JS challenge, this skill will automatically try accessible mirrors, extract the answer key from a mirrored PDF if needed, and cross-check the value against a second official source before reporting results.

Quick Start

Use the external-reference-verification skill to confirm the correct answer for question 12 from the blocked IRS publication webpage.