本地优先的个人 A 股研究决策工作台 · 案卷式研究闭环：进口判断→证伪→复盘→记忆。人做决策，AI 做广度/证伪/风险。A local-first personal A-share research OS.

Local-first AI research workbench for quant. Turn a natural-language strategy idea into a reproducible, auditable backtest — factor code, Reviewer checks, charts, and research notes, all saved locally.面向本地的量化AI研究工作台。将自然语言策略想法转化为可复现、可审计的回测结果——因子代码、审核员检查、图表和研究笔记，全部保存于本地。

Cognitive Intelligent Cooperation framework — human–AI co-creation architecture, state logic, governance boundaries. DOI: 10.5281/zenodo.20073579 · Active research workbench.

Messy long context -> structured insight -> auditable markdown, research threads, and inspectable outcomes.

Local-first asset-pricing and macro mispricing research system with alternative-data pipelines and research workflow orchestration.

Public sanitized release of my read-only crypto research workbench: frontend shell, Python JSON API, synthetic fixtures, tests, CI, and bilingual docs.

Reproducible corpus workbench testing Koenigsberg's Law of Sacrifice via Conceptual Metaphor Theory and LCC-style metaphor analysis across four historical cases: American Revolution, Napoleon, Lincoln, and Hitler. Quarto research site + manifest-driven Python pipeline.