Kalshi

**Judgement:** A strong, focused read-only integration for Kalshi prediction markets, well-suited for agents that need live event probabilities and basic portfolio analytics, but it exposes sensitive financial information and can easily drift into de facto investment advice if not carefully constrained. **What it does well** - Pulls live market data: trending markets, search by query, and detailed views of specific markets (prices, volume, orderbook depth). - Computes simple opportunity metrics using explicit expected-value logic and liquidity checks, helpful for ranking or filtering markets. - Supports authenticated portfolio views (positions, balance, history) via local API keys and RSA signing, enabling personalized analysis without trade execution. - Read-only access to Kalshi avoids direct trading or fund-loss risk. **Key risks / limitations** - **Financial privacy:** Portfolio and history endpoints reveal sensitive financial data; agents using this must avoid unnecessary data pulls and be prevented from exfiltrating credentials or position details. - **Investment-advice risk:** The built-in “opportunities” and trade-recommendation framing can lead agents to provide actionable trading advice; deployers should add strict policy and prompt guardrails around financial recommendations and risk disclosures. - **External dependency:** Reliability and latency depend on Kalshi’s API availability and rate limits; graceful degradation and error handling should be verified in practice. - **Scope:** Focused on data retrieval and simple EV-based screening; it does not provide advanced quantitative analytics or cross-platform arbitrage out of the box. **Recommended scenarios** - Agents that answer questions like “What are markets implying about X?” using up-to-date Kalshi prices. - Analytical or research tools that compare user beliefs to market odds, or summarize what prediction markets imply about elections, macro events, or policy outcomes. - Internal dashboards or expert-facing tools where users explicitly opt in to connecting their Kalshi portfolio for performance review and risk introspection (not for automated trading). - Educational or forecasting-assistant agents that use real market data to illustrate uncertainty, scenario probabilities, and how prediction markets respond to news, with clear disclaimers that outputs are not financial advice.