What problem does it solve? Financial answers built on stale or cached data can mislead users making time-sensitive decisions. This Skill defines clear rules for when the Hub may reuse session cache and when it must call a specialist for live market, prediction-market, or backtest data. ## Core Features & Use Cases - Cache vs. Specialist Routing: Decides whether a cached result directly answers the request or whether a specialist call is required for current prices, odds, news, or job state. - Numeric Claim Grounding: Ties every numeric claim to a specialist output or valid cache entry, preserving units, dates, and distinguishing current, historical, forecast, and backtested values. - Source Priority Ordering: Resolves conflicting sources by preferring fresh specialist output over session cache, runtime context, and generic model knowledge. - Use Case: When a user asks a follow-up about NVDA's current price after an earlier screener result, the Skill directs the Hub to call a specialist rather than present the stale cached quote as current. ## Quick Start Ask the Hub whether the cached prediction-market odds from earlier in this session are still fresh enough to answer my question, or whether it should fetch live data.
