ato_rate_scraping_skill

Overview

This skill automatically scrapes and extracts current Australian tax rates, thresholds, and benchmarks from ATO.gov.au and related government sources. It converts pages into structured rate data with provenance and basic validation. The output is designed for safe cache updates and audit logging.

How this skill works

The skill fetches ATO pages through a web-to-markdown reader API, preserving tables and key paragraphs. It locates known headers or table patterns, parses markdown tables into structured key-value pairs, and validates values against expected ranges. Each extraction records source URL, timestamp, financial year attribution, effective dates, and a confidence flag before returning a standardized rate object.

When to use it

• Validate cached tax rates against current ATO publications
• Detect quarterly changes such as fuel tax credit updates
• Confirm annual rate changes after budgets (SG, FBT, corporate tax, Div 7A)
• Populate rate tables for a new financial year or release
• Cross-check fallback rate values used by downstream analysis engines

Best practices

• Always validate extracted values against defined expected ranges before applying updates
• Flag and hold any values outside expected ranges for manual review — do not auto-update
• Log full provenance for every scrape: source URL, scrape timestamp, effective dates and FY attribution
• Respect site rate limits (suggested max 1 request per 10 seconds) and cache reader responses for short periods (e.g., 15 minutes)
• Keep resilient fallback values in code so scraping augments but does not replace hardcoded defaults

Example use cases

• Automated nightly job to refresh tax rate cache and alert on changes
• Quarterly monitoring of fuel tax credit tables and updating per-litre rates in billing systems
• Post-budget audit to detect SG, Medicare levy, or corporate tax changes and prepare releases
• CI test that ensures analytic models use the latest Division 7A benchmark interest rate
• Populate a web app’s rate table for the upcoming financial year with provenance metadata