Runner guide (advanced)

1. Evidence snapshot (Studio): JSON of what you configured and saw in the in-browser workspace; for audit, support, and reproducibility, not orders.
2. Strategy Runner Package v1 (Studio advanced download): deterministic JSON contract for your machine: strategy key, params, market context, non-custodial framing. QuantNest does not execute this file.
3. Starter-bot (local, open foundation): ships in the product repo under starter-bot/: you run it on your hardware; it validates the package, can fetch public Kraken OHLCV (no API keys), and runs bounded in-process paper simulation for registry ema_crossover v1 (EMA recurrence + basic crossover paper events). It is not a full backtest engine, full Studio parity, an uploaded-strategy runtime, or a live execution platform.
4. Local execution: the reference starter-bot client can run paper, dry-run, and guarded live modes on your machine when you install it yourself.QuantNest does not host this, does not hold your API keys, and does not place orders. There is no "go live" button in the web app.

From Studio, Export snapshot downloads a JSON Evidence snapshot of the workspace at that moment.

What it is

• Symbol, timeframe, parameters, and preview candle-window summary
• Optional latest backtest summary (counts, return, win rate, Sharpe, etc.) when you have run the engine
• References that help support or audit: share URL, strategy identifiers, marketplace slug when known
• A stable way to say “this is what I was looking at” when you email us or your own notes

What it is not

• Not live-trading ready: you cannot upload it to an exchange as orders
• Not the same JSON as Strategy Runner Package. That is a separate Studio download for tooling on your machine
• Not a promise that every strategy reproduces byte-for-byte outside QuantNest today

1. Attach the JSON file, or paste the top-level quantnest_export object from it, when you contact Support or support@quantnest.com.
2. Include the Studio Share link from the same session so we can open the same view in the app.
3. For your own records, store the file with a date and ticket ID; the filename encodes symbol, timeframe, and timestamp.
4. Treat it like evidence metadata, not a secret key file. It may contain public market context and IDs, not passwords.

In Studio, use Runner package for the v1 JSON contract aimed at local clients (including the starter-bot foundation in starter-bot/). It encodes strategy key, parameters, symbol/timeframe context, explicit non-custodial framing, and execution metadata (Kraken intent, paper-first recommendation). QuantNest does not run this file for you, does not hold funds, and does not place orders.

Current limitations

• Runner package export is limited to ema_crossover@v1 (semantics_version in JSON). Other strategies cannot be exported on this contract today.
• Studio blocks export when the symbol is unresolved or the timeframe is not on the starter-bot Kraken public OHLCV allow-list (e.g. 1mth). Use Evidence snapshot for full Studio context instead.
• No hosted execution, queues, or secrets handling from QuantNest. API keys stay on your hardware
• The starter-bot reference is separate from the web app: you run it locally; it validates the JSON and supports a narrow paper simulation today (ema_crossover v1 EMA recurrence + basic crossover paper events). Treat the download as a reproducibility and wiring pack, not a universal runtime or hosted automation switch.
• Prefer paper or read-only exchange modes first; never enable withdrawal permissions on automation keys

The canonical TypeScript schema and builder live in quantnest-frontend/src/lib/strategyRunnerPackage.ts; Studio wiring lives in studioRunnerPackageExport.ts.

Starter-bot (source)

The reference starter-bot tree lives in this product repository at starter-bot/ (README + contract docs describe v0.1 scope). An organization mirror may also exist at https://github.com/quantnesthq/starter-bot when published; use the in-repo tree as the default source of truth for what ships today.

We are not implying every marketplace strategy is executable outside QuantNest today. Coverage stays strategy-specific; this export is for reproducibility and local wiring experiments, not a universal runtime guarantee.

Studio offers two downloads. They are not interchangeable.

• Evidence snapshot (for you, support, and audit): what you configured and saw in Studio (including optional backtest summary). Not an order file.
• Runner package (for software on your machine): strategy key, params, normalized_pair, quote currency, and non-custodial framing. Consumed by clients such as starter-bot.

Frozen ema_crossover@v1 signal math does not guarantee identical candles: in-app backtest history and Kraken OHLCV on your machine can differ. Compare outcomes; do not assume pixel-perfect parity.