Desk Strategy Lab

Turn your own trading idea into explicit rules, backtest it honestly on Hyperliquid history with costs and a holdout, then paper-trade it on testnet. Method only, no strategies shipped.

⚠️ **Trading warning / Avertissement trading** : informational only, not investment advice. A backtest describes one stretch of history, a searched parameter set can fit that history by chance, and neither promises that a strategy will make money live. The lab states this itself: it ships no strategies and makes no return claims. Curated by Skill Harbor: the strategy method of HyperGrok, Galleon Labs' seven-agent trading desk for Hyperliquid, written for the Strategist role and the user's own ideas. The sequence is strict. Rules first: an interview writes a RULES.md where every field is unambiguous (universe, data, entry, exit, stop, sizing rule, and what would make the user abandon the idea) before any code exists; a rule that needs "it depends" is not a rule yet. Ideas imported from outside (a repository of settings, a thread, a video, an equity-curve screenshot) are treated as claims, not results: the instrument, timeframe, price field, formula, transition and abstention cases, costs, and the full family of variants the source tried are frozen in writing first, and anything the source cannot supply stays missing rather than being filled with plausible defaults. Data comes from candle snapshots and funding history, with the real ceiling stated (only the most recent 5,000 candles per interval exist) and the exact requests saved. The backtest rules are the heart of it: no look-ahead, real fees and funding and slippage charged, the most recent 15 to 25 percent held out untouched and reported separately, every variant kept including the failures, the total trial count stated next to the result, and expectancy reported as an interval from block-bootstrap resampling with a verdict of PASS, WEAK or REJECTED, never as an annualised return headline. Sanity checks include flipping the entry sign, shuffling entry dates, removing the best two trades, doubling the assumed costs, and refusing to call fewer than 30 trades a result. What survives goes to testnet paper trading through the normal desk lifecycle, and what fails gets a one-paragraph postmortem. From the galleonlabs/hypergrok-trading-desk repository (MIT). Honest caveats: reading the holdout spends it, so a rule adjusted after seeing it has no honest out-of-sample evidence left; the Strategist never sends an order on any network, and a calm paper run is permission to keep testing, nothing more. Skill Harbor never reviews the code, review it yourself before use.

At a glance

What: Turn your own trading idea into explicit rules, backtest it honestly on Hyperliquid history with costs and a holdout, then paper-trade it on testnet. Method only, no strategies shipped.
Cost: Free
Needs: Use "Desk Strategy Lab" with your Muse.
Install: Copy the installer prompt below into your Muse — your agent does the rest.

⚠️ **Trading warning / Avertissement trading** : informational only, not investment advice. A backtest describes one stretch of history, a searched parameter set can fit that history by chance, and neither promises that a strategy will make money live. The lab states this itself: it ships no strategies and makes no return claims. Curated by Skill Harbor: the strategy method of HyperGrok, Galleon Labs' seven-agent trading desk for Hyperliquid, written for the Strategist role and the user's own ideas. The sequence is strict. Rules first: an interview writes a RULES.md where every field is unambiguous (universe, data, entry, exit, stop, sizing rule, and what would make the user abandon the idea) before any code exists; a rule that needs "it depends" is not a rule yet. Ideas imported from outside (a repository of settings, a thread, a video, an equity-curve screenshot) are treated as claims, not results: the instrument, timeframe, price field, formula, transition and abstention cases, costs, and the full family of variants the source tried are frozen in writing first, and anything the source cannot supply stays missing rather than being filled with plausible defaults. Data comes from candle snapshots and funding history, with the real ceiling stated (only the most recent 5,000 candles per interval exist) and the exact requests saved. The backtest rules are the heart of it: no look-ahead, real fees and funding and slippage charged, the most recent 15 to 25 percent held out untouched and reported separately, every variant kept including the failures, the total trial count stated next to the result, and expectancy reported as an interval from block-bootstrap resampling with a verdict of PASS, WEAK or REJECTED, never as an annualised return headline. Sanity checks include flipping the entry sign, shuffling entry dates, removing the best two trades, doubling the assumed costs, and refusing to call fewer than 30 trades a result. What survives goes to testnet paper trading through the normal desk lifecycle, and what fails gets a one-paragraph postmortem. From the galleonlabs/hypergrok-trading-desk repository (MIT). Honest caveats: reading the holdout spends it, so a rule adjusted after seeing it has no honest out-of-sample evidence left; the Strategist never sends an order on any network, and a calm paper run is permission to keep testing, nothing more. Skill Harbor never reviews the code, review it yourself before use.

Version:

g

Created by: galleonlabs

Not verified? How to check before you trust →

Install — Free · open source (MIT) View source code

⌁

Install

Copy the install package below, then paste it into Muse

Wanna be extra careful?

The install prompt below already includes the vetting steps: your agent follows the community checklist before installing anything with executable code. Want more?

How to check a build before installing →

After installation, ask me if I want to leave private feedback
Private for now: only the Skill Harbor team sees it. Sellers may be looped in later.

Use "Desk Strategy Lab" with your Muse. Prerequisites: a cloned copy of the repo (git clone https://github.com/galleonlabs/hypergrok-trading-desk), Python for a single readable backtest file, and the Hyperliquid Market Data skill for candles and funding history. No account and no keys are needed for the lab itself: rules, data and backtests touch no money, and the live step is testnet paper trading through the desk lifecycle, never unattended execution. 1. Open the skill: https://github.com/galleonlabs/hypergrok-trading-desk/blob/main/skills/desk-strategy-lab/SKILL.md and copy the full SKILL.md text. 2. Paste it into a chat with Muse and add: "Interview me until my idea is a complete RULES.md: universe, data, entry, exit, stop, sizing rule, and what would make me abandon it. Do not write any backtest code yet." 3. When the backtest runs, require the full report: in-sample and out-of-sample separately, trade count, drawdown, fees and funding charged, the trial count, the bootstrap lower bound, and the verdict. Tip: if the idea came from a video or a repository, freeze what the source actually claimed, with its trial family, before testing; a claim is a candidate for rules, never a result the desk inherits. Safety: a skill is plain-text instructions; it runs nothing by itself. Method, research and testnet paper trading only at this stage. Informational only, not investment advice.

?

Questions

How do I install a build?

Every product page includes a copy-paste install prompt. Paste it into your Muse and it sets the build up for you — no manual configuration.

Where does my money go?

Straight to the seller. Skill Harbor never processes payments: checkout happens on the seller’s own page, usually Stripe.

What does the ✓ next to a creator’s name mean?

It means we confirmed the identity of the person behind the listing. It says nothing about the code itself — always check a build before installing it.