Your First Script in Furge

Run a read-only Python script, understand its output, and follow it through deployment.

1. Open an example

Open Scripts in Furge, sign in, and choose Examples. Select Watch candle closes, then Use example. This creates a separate script; it never replaces an existing one.

You do not need to install Python or fund an exchange account to write this example. Running it requires a ready Furge runtime. If your runtime is stopped, start it from Runtime first.

script = {"name": "candle-monitor", "version": "2", "lookback": 2}
SOURCE = "btc@candles@hyperliquidf:timeframe=60"


def on_data(ctx, history):
    candle = history.source(SOURCE, 0)
    if candle is None:
        return
    print(f"BTC close: {candle['c']} | volume: {candle['volume']}")
    return {"metrics": {"close": candle["c"], "volume": candle["volume"]}}

This script only reads data and prints output. It contains no order or swap calls.

2. Understand the source

btc@candles@hyperliquidf:timeframe=60 means BTC candles from Hyperliquid with a 60-second interval. It is not an execution instruction.

history.source(SOURCE, 0) returns the newest retained candle, or None before one is available. Read Candles for the complete returned dictionary. Use Data sources in the editor to find the other source types and exchange support.

Edits save automatically. Wait for Saved before leaving; starting a run also saves the current revision.

3. Run a historical backtest

In Run settings, choose a recent completed UTC range under Historical backtest, then click Backtest. No trading wallet or real funds are used.

The Backtests panel below the editor shows progress. When the run completes:

  • Results: the latest close and volume metrics. Zero closed trades is correct for this read-only script.
  • Logs: one printed line per processed candle, such as BTC close: 80500.0 | volume: 12.4 (illustrative values).
  • Source: the exact saved Python revision used by that run.
  • Artifacts: files explicitly produced by your script; this example creates none.

If no records arrive, check the source's historical support and date range. A live-only source cannot be backtested, even when the terminal chart has history.

4. Observe it live

Set a short Duration under Bot deployment, then click Deploy bot. Review the confirmation and deploy. The same read-only example still contains no trading instructions; deployment itself is not an instruction to buy.

Furge opens the job page. Watch Results for events or Logs for Python output. A 60-second candle appears only after a complete interval with trades; the first partial interval is discarded. A quiet feed does not necessarily mean a failed job.

Click Stop to end the job. Closing the editor or browser does not stop it. Reopen it from Scripts → Jobs.

5. Add trading deliberately

Use Execution to choose supported actions for your exchange. Furge scripts use configured automated trading accounts, not your browser wallet. For Montra, fund the Elysium Automated Trading Wallet with the input asset and HYPE for gas.

Backtest supported strategies before enabling real execution. Montra and Nest contract actions run live. Stopping a job prevents future requests but does not cancel resting orders or reverse transactions already submitted.

Next: Examples · Data Sources · Monitoring and Troubleshooting.

On this page