Releases $30k open-source trading terminal

A May 21 YouTube video from Bitwise AI put fresh attention on HKUDS’s AI-Trader, an open-source project described in its GitHub README as “100% Fully-Automated Agent-Native Trading.” The video’s description called it a “$30k AI trading terminal,” said it had 21,838 GitHub stars, 37 AI agents and a full Model Context Protocol, or MCP, host, and framed it as a free alternative to a Bloomberg Terminal. (youtube.com) GitHub showed the repository at 18.4k stars on May 22, with 2.8k forks and 343 commits, indicating the count was moving quickly or that the video used a more recent snapshot than the crawler captured. (github.com) The project’s README says it is built so “AI agents need their own” trading platform, not just human users. ### Why did this repository get attention beyond another trading bot? The video description emphasized packaging over novelty: 37 agents, an MCP host and a complete interface rather than a narrow proof of concept. That matters because the project is presented as a full operating surface for research, market intelligence and execution workflows, not just a single strategy script. (youtube.com) HKUDS’s README uses the same framing. It says “Just like humans have their trading platforms, AI agents need their own,” and describes AI-Trader as an “Agent-Native Trading Platform” where agents can join by reading a skill file and registering. (github.com) ### What is actually in the codebase? GitHub’s public page says AI-Trader supports “all major AI agents, including OpenClaw, nanobot, Claude Code, Codex, Cursor, and more.” The README also points to a live trading platform and lists recent updates including capacity and worker-throttling upgrades on May 12, experiment notice exposure tracking on May 13, and earlier production stability work in April. (youtube.com) The repository says its FastAPI web service runs separately from background workers so user-facing pages remain responsive while prices, profit history, settlements and market-intelligence jobs run out of band. (github.com) On March 21, according to the README, the project added a dashboard page described as a “unified control center for all trading insights,” and on March 3 it launched Polymarket paper trading with real market data and simulated execution. ### Where does the “$30k terminal” claim come from? Bitwise AI used the “$30k AI trading terminal” language in the YouTube title and description, but the available public GitHub page does not show a line-item valuation for the software stack. (github.com) The wording appears to be the video creator’s characterization of the product’s scope and comparable value, rather than a price posted by HKUDS on the repository page. The clearer public claim from HKUDS is functional rather than financial: the repository is pitched as infrastructure for autonomous or semi-autonomous trading agents, with live services, dashboards and paper-trading support already visible in the README. (github.com) ### Why are developers focusing on products like this? GitHub’s repository page shows AI-Trader as more than a model wrapper. The public materials highlight data feeds, dashboards, market-intelligence jobs, background processing and agent registration, which are the kinds of workflow pieces developers usually need to combine before a system looks deployable. (youtube.com) The video’s framing made the same point by stressing the assembled product — not just the model count — in its headline numbers. In practice, the attention around AI-Trader suggests developers are rewarding open-source projects that package ingestion, summarization, interfaces and automation into a recognizable end product. (github.com) That last point is an inference from the project’s feature set and the way the video presented it. ### What comes next for the project? HKUDS recorded a README update two days before the May 22 GitHub crawl, and the latest listed code commit was “yesterday,” indicating active development. (github.com) The repository directs users to ai4trade.ai for the live platform and to a SKILL.md registration flow for agents joining the system. (youtube.com)