Give your test market buyers and sellers who use the app.

A trading interface needs more than a conversation between simulated traders. It needs accounts that select a side, enter a price and quantity, submit an order and inspect the result. Jango gives cast members browser controls to perform those steps in a test app.

Set up your test participants

Assign a different goal to each account

Create buyer and seller accounts in your sandbox. Give the buyer a limit price and quantity, and give the seller an offer that can match. Set each participant’s starting URL to the order-entry screen and explain the relevant controls. Participants act through the interface; your app owns the order book and matching logic.

Select Buy or Sell and fill the price and quantity fields.
Submit a test order and inspect the account’s open orders.
Place a compatible order from another account and compare visible fills.
Ask a participant to cancel a remaining order when the app exposes that action.

A concrete first scenario

In a disposable test market, ask one account to buy 3 units at a limit of $100 and another to sell 2 at $99. Inspect the resulting trade and remaining quantity in your app. These inputs create a useful partial-fill scenario; your matching rules determine the execution price. Jango’s included browser fixture has an automated test for separate identities placing orders, partial fills and prevention of self-trades.

The same approach works for marketplace forms

Assign a seller to create a test listing and a buyer to find it, inspect its details and take an available test action. Browser actions are not tied to a chat template. Use specific goals, development accounts and visible form controls, then inspect the state from both accounts.

What is and is not verified

The included order-book test exercises real browser controls with scripted actions. It does not establish that an AI model will independently complete every exchange’s interface. Validate your own app and model combination. Jango is for sandbox interaction testing, not live trading, financial market prediction, high-frequency load generation or proof that a matching engine is correct. Keep deterministic matching, balance and concurrency tests alongside these runs.