prinzclaw-scan

name	prinzclaw-scan
description	Scans public statements for verifiable promises with time anchors and numeric metrics
tools	["web_fetch","rss_parser"]

Scan Skill — T (Tension)

Purpose

Extract verifiable promises from public statements by powerful figures.

When to Use

When given a quote or statement from a public figure, use this skill to determine if it contains a verifiable promise.

How It Works

Input

• Entity name and handle
• Quote/statement text
• Source URL (optional)
• Timestamp (defaults to now)

Analysis Criteria

A promise is verifiable if it contains at least TWO of:

1. Time anchor: "within 6 months", "by 2026", "next quarter", "in 30 days"
2. Numeric metric: "$2", "50,000 jobs", "15% growth", "reduce to $X"
3. Binary outcome: "will not sign", "will never do", "I guarantee X won't happen"

Filtering Rules

Reject statements that are:

• Vague slogans without specifics
• Political rhetoric without measurable outcomes
• Emotional expressions without facts
• Pure opinions without evidence

Output

Returns a structured promise object:

{
  "id": "SHA-256 hash",
  "entity": "Name",
  "promise_text": "exact quote",
  "source_url": "https://...",
  "promise_date": "2026-03-16",
  "expiry_date": "2026-09-16",
  "verifiable_metric": "price < $2.00",
  "data_source_for_verification": "https://eia.gov/...",
  "status": "PENDING"
}

Usage in TACO Pipeline

The scan skill is the first stage. Its output feeds directly into the analyze skill for contradiction detection.