A while back I came across TradingAgents — a really cool multi-agent LLM stock analysis framework where like a dozen "agents" (market analyst, news analyst, bull researcher, bear researcher, risk team, etc.) debate a stock and produce a final trade recommendation. The output is genuinely interesting to read.

Problem: it ships as a CLI. You pick options in a terminal, watch logs scroll, then go hunt for markdown files on disk. The reports are good, the experience of getting to them isn't.

So I forked it and bolted on a web GUI. Runs locally, talks to whatever LLM provider you have a key for (OpenAI, Anthropic, Google, OpenRouter, DeepSeek, Ollama, xAI, Qwen, GLM, MiniMax). All Apache 2.0.

Some things I ended up adding because I wanted them:

• Live pipeline visualization showing which agent is working
• Reports tab with a 3-pane reader, table-of-contents, search
• A "report length" knob (Concise / Standard / Comprehensive) — concise mode saves \~50% tokens
• Multi-session chat where you can pin past reports as grounding context and ask follow-up questions
• Three themes because I couldn't decide

Sample reports:

• AAPL
• NVDA

Repo: https://github.com/TheLocalLab/TradingAgents-GUI