blacksnow
от sieershafilone
Search & ResearchКратко
Detects pre-news ambient risk signals across human, legal.
Описание
Что делает этот навык
BlackSnow ingests fragmented, legally accessible data exhaust from procurement, regulatory, labor, and logistics sources. It applies Bayesian evidence accumulation and ontology alignment to surface pre-event risk vectors before formal news or disclosures occur. Outputs are structured risk primitives consumable by financial, insurance, logistics, and policy systems.
Surfaces risk signals weeks before formal events by correlating individually weak public data points that news monitoring and sentiment analysis miss.
Когда его использовать
- Detecting infrastructure stress signals before grid outages
- Flagging supply chain disruptions ahead of formal announcements
- Identifying regulatory drift before rules are published
- Generating risk primitives for insurance underwriting decisions
- Monitoring labor system stress indicators before strikes or freezes
Пример сценария
User asks: What energy infrastructure risk signals exist in a target region over the next 30 days?
- 1
Harvester collects public procurement notices, maintenance deferrals, and labor filings from approved domains
- 2
Normalizer maps heterogeneous inputs into a unified risk ontology
- 3
Accumulator performs Bayesian evidence accumulation across contributing domains over time
- 4
Forecaster estimates outcome likelihoods and temporal windows
- 5
Packager outputs a structured risk primitive with confidence score and tradability flags by market type
JSON risk vector with signal confidence, 21-45 day time horizon, contributing domains, likely outcomes (localized outage, price volatility, policy intervention), and tradability flags for insurance and commodities
Требования
- BlackSnow subscription (Observer tier starts at $99/mo, Operator at $1,500/mo, Fund/API at $10k-50k/mo)
- Gated onboarding approval and completed audit before deployment access
Установка
clawhub install sieershafilone/blacksnowВыполните команду CLI или вставьте промпт в Claude Code, Cursor, Codex — любой кодинг-агент, которым пользуетесь.