Show HN: PrimerKit, financial market work at agent speed

An open-source project called PrimerKit launched on Hacker News as a workspace where AI agents handle financial market research, planning, testing and order execution, with every run logged in a local journal and each order requiring user approval. PrimerKit defaults to PreReason as its data source, connects to hosted or local models including Ollama and LM Studio, and keeps brokers disabled until the user enables them via /connect. The tool states it never proposes a strategy and that live orders carry real money and real losses.

Made to be operated, not only seen. primerkit is the environment your AI agent works in for the mechanics of financial data: research, insight, planning, testing and action. The checks stay in: answers checked against the data, every run in a journal on your machine, and your yes asked before each order an exit you set goes on the yes to its purchase . From research outward to action: a paper order, an order you approve, a chart on your screen. Nothing here is investment advice: the model never proposes a strategy, and answers from the data or applies rules you wrote yourself. A live order is real money, and its losses are real. PROpen source. Default data source: PreReason https://www.prereason.com . CONNECTIONS - ●modelan AI app you already use, a hosted model on your own key, or one on your machine Ollama or LM Studio - ●PreReasondefault data source - ○brokersoff until you turn one on: /connect FUNCTIONSa code and Enter runs it - DG/digestwhat changed - CP/comparebriefings and web search - BR/briefa briefing as a board - JR/journalevery run, kept on your machine - CH/charta chart - MD/modelswitch the model - CT/catalogevery briefing, metric - CN/connectdata and brokers