Agent Skills › Superior-Trade/superior-skills › intelligence

intelligence

GitHub

用于查询市场热点、资产信号及单标的深度分析，调用 Unified API 获取实时数据与证据。涵盖扫描、回测至部署流程，强调基于真实数据而非训练数据提供结果。

skills/intelligence/SKILL.md Superior-Trade/superior-skills

触发场景

询问市场热门趋势或动态查询特定符号的市场情报进行市场扫描或单标的钻取分析

安装

npx skills add Superior-Trade/superior-skills --skill intelligence -g -y

更多选项

不安装直接使用

npx skills use Superior-Trade/superior-skills@intelligence

指定 Agent (Claude Code)

npx skills add Superior-Trade/superior-skills --skill intelligence -a claude-code -g -y

安装 repo 全部 skill

npx skills add Superior-Trade/superior-skills --all -g -y

预览 repo 内 skill

npx skills add Superior-Trade/superior-skills --list

SKILL.md

Frontmatter

{
    "name": "intelligence",
    "metadata": {
        "updated": 1788393600,
        "version": "0.2.0"
    },
    "description": "Use when the user asks what's hot, what's moving, if there's any alpha, what's the setup on a symbol, or any market-scan or single-symbol drilldown question. Uses Superior Trade Unified context data and its evidence fields rather than stale training data."
}

Intelligence

Use Unified API market intelligence to rank current setups and inspect one canonical symbol. Read the returned evidence; never treat a score as advice.

Files in this skill

File	What it covers
`references/api.md`	Unified scan/setup parameters and contract-shaped examples.
`references/workflow.md`	Scan → setup → backtest → deploy workflows.
`references/buckets.md`	Optional strategy lenses for interpreting evidence; these are not Unified response categories.
`references/glossary.md`	How to explain common evidence fields without assuming they are always present.

Route selection

List/ranking question → GET /context/scan. Use bucket=all for an open-ended scan, or alts/hip3 when the user narrows the universe. Use category=all unless they explicitly request momentum, mean_reversion, breakout, or volume. top_n defaults to 10 and must not exceed 50.
Single-symbol question → first resolve a canonical symbol with GET /context/markets, then call GET /context/setup/{symbol} with the URL-encoded canonical value.

Presenting results

State computed_at when the response includes it.
Present the returned symbol, category, and score without renaming them to an older bucket taxonomy.
Cite the exact evidence values returned for each conclusion.
Treat setup fields such as regime, levels, volatility, and funding as optional unless they appear in the response.
Before recommending capital, apply an appropriate strategy lens, run a backtest, and explain the principal risk.

Never manufacture an empty or unavailable scan from training data. Scores are deterministic market computations, not guarantees or a signal service.