Open-source LangAlpha launches as the Claude Code of finance for natural-language investment research

Open-source project LangAlpha has officially launched as a natural-language-driven investment research agent, with InfoQ describing it as the Claude Code of finance.

The project's GitHub repository, ginlix-ai/LangAlpha, describes itself as Claude Code for Financial Market, is licensed under Apache 2.0, and is built on LangChain with support for Python 3.13 and above.

LangAlpha's core pitch is moving from vibe coding to vibe investing. It argues that existing AI finance tools treat investing as a one-shot Q&A — ask a question, get an answer, move on — while real investing is a Bayesian, iterative process in which theses are refined over weeks and months as new data arrives.

Borrowing the pattern that made code agents like Claude Code and OpenCode successful, LangAlpha gives each research goal (such as a Q2 rebalance or a data center demand deep dive) a persistent workspace: the agent interviews the user about goals and style, produces its first deliverable, and saves files, threads, and accumulated research for the next session.

On the technical side, LangAlpha ships parallel research subagents for market screening, and a Programmatic Tool Calling (PTC) mode in which the agent writes and executes Python inside a cloud sandbox to process financial data, rather than pouring raw data into the LLM context window — cutting token waste while enabling multi-step analysis.

For data, it runs a three-tier provider fallback chain — the hosted ginlix-data proxy, FMP, and free Yahoo Finance — and ships 23 pre-built financial research skills covering DCF models, initiating coverage, earnings analysis, and morning notes. The model layer is bring-your-own-key, with OAuth support for ChatGPT and Claude subscriptions as well as coding plans from Kimi, GLM, MiniMax, and Doubao, plus automatic failover.

A web workbench offers TradingView charting and a source-provenance panel, and the agent can be used from Slack, Discord, Feishu, and Telegram, with scheduled or price-triggered automations.

By open-sourcing an institutional-grade investment research workflow, LangAlpha lowers the barrier for individual developers and small teams; the questions to watch are how quickly it gains traction in the community and how these autonomous agents hold up in compliance-sensitive financial settings.