What problem does it solve? Montréal's open datasets update at wildly different rates — BIXI station status refreshes every minute while property assessments update every three years — and agents risk presenting outdated data as current without a systematic way to check freshness. ## Core Features & Use Cases - Update Frequency Reference: Categorized tables of update rates (real-time, daily, monthly, annual) for major datasets like BIXI, 311 requests, building permits, and crime data. - Freshness Detection Methods: Three concrete techniques — CKAN package metadata timestamps, MAX() queries on date fields via DataStore SQL, and GBFS last_updated timestamps. - Staleness Thresholds & Seasonal Awareness: Rules of thumb for when data counts as stale, plus seasonal behavior for BIXI, skating rinks, and snow removal so agents don't misreport off-season zeros as errors. - Use Case: A user asks "How many bikes are at the BIXI station near me?" in January. The agent recognizes BIXI is seasonal (April–November) and explains the closure instead of reporting zero available bikes as a live fact. ## Quick Start Ask the agent when the building permits dataset was last updated and whether the data is still fresh enough to trust.
