Minara AI has released its new generation of general-purpose financial intelligence agent, Minara Harness, available for global open download. Features include 7×24 investment research and analysis across all asset classes, writing reports in natural language and quantitative strategies, analyzing investment opportunities across multiple dimensions with multiple specialized Agents, choosing underlying large language models, and even subscribing to and directly running trading strategies written by others, with support for post-hoc decision review.
Official benchmark tests claim it outperformed OpenClaw, Hermes, and Claude Code on financial report, web research, and spreadsheet tasks—this result is currently reported only by Minara itself. We haven’t found any third-party independent reproductions, so treat it as an official self-test rather than an independently verified conclusion.
Worth noting is the "automated trading within authorized scope" feature: the strategy may be written by someone else, the model may be selected by you, but the final profits and losses are yours. Before you proceed, clarify the authorized scope, the stop-loss mechanism, and the real historical performance of the other party’s strategy—don’t hand over your positions just because the operations feel smooth.
Would you be willing to delegate trading authorization to a general financial agent, or would you prefer to use it only for investment research? Is there a market for this kind of future?
#Minara #AIAgent #自动化交易 #金融智能体