sqx-labのご紹介：StrategyQuant向けのAIオーサリングツールキットが、無料でダウンロード可能になりました

In July I wrote about building StrategyQuant blocks, groups and templates with AI — four AI skills that turn plain English into import-ready StrategyQuant artifacts, each one validated against your own installation before it ever reaches you. At the time, the toolkit was something I demonstrated. Today it is something you can install.

sqx-lab — the SQX Authoring Toolkit — is now packaged as a single Claude Code plugin, free to download from the StrategyQuant Codebase.

What you get

Four skills, covering the whole authoring chain that feeds the StrategyQuant builder:

• sqx-custom-block — describe an atomic trading rule (“close above the upper Bollinger band, with its opposite”) and get validated custom-block XML, built only from indicators your install actually has.
• sqx-random-group — pool your blocks into the menus the builder samples from: Condition groups for rules, Value groups for prices and levels.
• sqx-strategy-template — describe a thesis (“breakouts only in the trend direction, entered on a stop at the prior-period high”) and get a buildable .sqx template wired to your clean groups, with a full exit stack.
• sqx-strategy-project — wrap templates into a runnable project.cfx with build tasks, data, timeframes and analysis wired in.

The chain runs block → group → template → project. Each output feeds the next, or any skill works alone. And the principle behind all four is unchanged: the AI proposes, but your StrategyQuant install disposes. Every artifact is derived from a catalog of what your build actually supports and validated before hand-off — no phantom indicators, no hand-written XML, no plausible-looking files that quietly refuse to build.

New since the July article

The release version (1.2) is built around a simple idea: setup should be one command, and failure should be loud.

• /sqx-setup

  — point it at your StrategyQuant X folder once. It validates the folder (and refuses anything that isn’t a real SQX install), stores the path for all four skills, bootstraps every catalog, and reports what it found. A wrong folder can no longer silently break the toolkit.

• /sqx-doctor

  — a one-command health check: Python, install, catalogs, and whether the block → group → template → project chain is actually intact. It specifically hunts ランダムなグループの分割に失敗しました — groups referencing custom blocks that aren’t in your install. One broken Value group used to silently disable entire classes of strategy shapes; now the doctor names it, lists the missing blocks, and the Custom Block Builder’s Repair mode rebuilds them.
• Settings survive updates — your install path lives outside the plugin now, so updating is: download, reinstall, re-run

  /sqx-setup

  . Seconds.

What you need

• StrategyQuant X (build 144) — your own install; everything is derived from it
• Claude Code — the free plugin installs into it
• Python 3.8+ — standard library only, no pip installs

Everything runs locally. Your strategies, catalogs and data never leave your machine.

Get it

Download the zip, unzip, and install inside Claude Code:

/plugin marketplace add /absolute/path/to/sqx-lab
/plugin install sqx-lab@sqx-lab
/sqx-setup

That’s the whole setup. From then on you don’t invoke skills by name — you just describe the artifact you want, and Claude picks the right layer.

Download the SQX Authoring Toolkit from the Codebase →

The full manual — six-step tutorial, per-skill reference, FAQ — lives on the Codebase entry alongside the download.

Ivan Hudec (clonex) has published 135+ free extensions in the StrategyQuant Codebase and writes about the research behind them at algotrading.space.