Superior Terminal Is Now Open Source

We just open-sourced our trading terminal. Everything a person touches is in the repo: the chart you draw on, the agent that reads the drawing, the validator that refuses unsafe strategies, the deploy pipeline. Apache-2.0, on GitHub, today. Two keys and it runs on your machine.
A trading tool asks for your API keys and then touches real money with them. You should get to read the thing you're trusting. Now you can — every prompt and safety rule in the pipeline is in the repo, and all of it is yours to edit.
What it does
Draw your read
Sketch the level or zone you keep staring at. The agent reads your drawing against the live chart and answers with a concrete plan: entry, stop, target, invalidation. If you'd rather type than draw, that works too.
Recurring strategies
Describe a rule you want running on the market and the terminal writes it as a real Freqtrade or Nautilus-Trader strategy, tested against up to two years of Hyperliquid data before it deploys with live PnL tracking. The output is plain Freqtrade code.
One-shot setups
For a single trade idea, the agent places a bracket order with your entry, stop and target, plus an auto-stop timer. You size it with a slider. The agent never sizes a trade for you.
A validator with scars
Before anything deploys, the strategy is checked for the failure modes that kill homebrew bots: lookahead bias, stops that ignore leverage, entries firing on dead candles, indicators still warming up, off-by-one TA-Lib indexing. Every rule exists because a strategy once reached production without it. The checks are in the repo.
What open means here
The terminal is the interface to the same infrastructure that runs terminal.superior.trade. Self-host it and your keys never leave home. Use the hosted build and skip the setup. Same API, same accounts. The execution core stays closed. Everything that handles your keys is in the repo.
If your trading happens through an AI agent, hand it one line: Read superior.trade/SKILL.md and register. Claude Code, OpenClaw, Codex and Cursor all speak it.
In March 2025 we let autonomous agents trade real money for 36 days. One returned 34.8%. Others crashed. We're not promising you that run. The failures all happened in the same place: the execution path between an idea and a filled order. That path is now open source.
From idea to live trade in minutes.
Describe a strategy in plain language and get validated, backtested code running live.
Open Terminal