jesson-hh / financial-analyst

What is financial-analyst?

A Python-based A-share research workstation that orchestrates 24 AI agents to produce deep-dive stock research reports from a single command. You hand it a stock code like `fa report SH600519` and roughly ten minutes later get a markdown report covering fundamentals, technicals, whale behavior, quant model scores, and a bull/bear/risk debate -- all written by specialized agents working in four trust tiers. The system integrates with major LLM providers (qwen, deepseek, openai, anthropic) and exposes 20 MCP tools so it can be called directly from AI IDEs like Claude Desktop, Cursor, or JetBrains AI plugins.

Why is it gaining traction?

The hook is turning a complex multi-agent workflow into a one-liner: no prompt engineering, no stitching together separate tools, no infrastructure to maintain. The four-tier trust architecture handles the messy reality of Chinese financial data -- untrusted news feeds are JSON-schema-locked at Tier-1, only the report-writer agent can write files, and everything feeds into a bull/bear/risk debate before synthesis. The pluggable memory system means you can edit a markdown file and the next report immediately reflects your custom rules, without restarting or redeploying. For developers already working in the Chinese market, this bridges the gap between generic LLM tooling and domain-specific research workflows.

Who should use this?

Quant researchers and retail investors focused on the Chinese A-share market who want structured, multi-perspective analysis without building it themselves. Financial analysts who need to compare multiple stocks quickly. Developers building tools that need domain-aware financial data extraction. It is less useful for US/European market analysis or anyone without experience navigating Chinese market infrastructure (Tushare tokens, TDX data feeds, domestic LLM providers).

Verdict

With 712 passing tests, active development (v1.0.7 released 2026-05-26), and 440 alpha factors built in, the engineering is serious. However, the 1.0% credibility score and 15 stars reflect a young, single-maintainer project with limited community exposure -- an early-adopter bet rather than a proven tool. If you work in A-share markets and want an open-source research workstation with AI agent orchestration, it is worth a closer look. If you need stability guarantees or vendor backing, wait for broader adoption.