Anyone doing investment research has a bit of a data hygiene problem: strategies, positions, and stock-selection logic are the most private things you own — the last thing you want is them living on someone else's server.
HunterCode (Community Edition, 521 stars, Apache-2.0, Python) is an open-source answer. Its pitch: a self-hosted AI investment research assistant — your data and your conversations stay on your machine.
What it is
A multi-agent investment research terminal covering A-shares, HK, and US stocks. The core word is local:
- Inference runs on your machine — not shipped to a cloud model;
- Positions never leave your disk — holdings, strategies, research notes stay local;
- BYOK — bring your own model keys, you keep control of the data;
- docker compose self-hosting — one command to bring the service up.
For people who do this seriously, it means: you get AI research capability without giving up data sovereignty.
Three signals it's worth watching
- "Local-first" hits a hard requirement in investing. The sensitivity of financial data makes "data never leaves the machine" a must, not a nice-to-have. Whoever makes on-prem AI research work well solves the compliance and privacy headache that private funds and professional individual investors feel most acutely.
- Multi-agent architecture. Not a single chatbot — multiple agents each owning a slice of the research workflow. That's closer to how real investment research actually gets done than "one big model answering questions."
- Pure open source, pure local. Apache-2.0, docker compose, BYOK — no cloud lock-in, no data exfil. The posture a serious research tool should have.
The honest caveat
It's for people comfortable tinkering, not a turnkey SaaS. You need to:
- be comfortable with docker compose deployment;
- have model API access (BYOK);
- sort out your own data sources (quotes, financials, etc.).
It's friendliest to private-fund research teams and willing individual investors — the "just install an app" crowd will hit a wall. Note too it positions itself as an open-source alternative to Tencent's WorkBuddy Finance Edition, with no affiliation to Tencent.
I've localized the README and core docs to Chinese: https://github.com/yangshun2005/hunter-community-cn
If you find this project useful, a star on the original repo supports the author's ongoing maintenance.
Top comments (0)