QuantMind is an agent-native knowledge extraction and retrieval framework for quantitative finance that transforms raw financial information into structured, typed knowledge.
The tool addresses the challenge of converting unstructured financial sources—research papers, news articles, regulatory filings—into reliable, queryable knowledge that downstream systems can act upon. It does this by refining raw information through deterministic verification pipelines while preserving citations and timestamps. Each piece of extracted knowledge is typed and remains independently queryable, enabling time-based searches and persistent storage. The framework is designed with agent-oriented workflows in mind: developers describe the pipeline they need, and coding agents can construct it within the repository against defined contracts, skills, and verification rules. The tool also functions as a standard Python library for direct integration.
The project suits teams building quantitative research workflows that need to ingest and structure financial data at scale. It is particularly relevant for organizations where coding agents handle pipeline construction alongside human oversight. The framework ships with initial implementations including PaperFlow and news collection capabilities, with additional flows planned. This approach differs from generic knowledge extraction tools by being purpose-built for quantitative finance, where typed knowledge, citation tracking, and temporal awareness are essential.
Development activity shows the project undergoing significant architectural changes to strengthen agent-native capabilities, with work focused on contexts, skills, and hooks that enable agents to perform high-quality work directly within the repository. The project has gained academic recognition through acceptance at a major workshop venue. The team is actively building out the initial release on the public repository and actively soliciting community participation.
