yaorenliang

Trading Agents Skill — Data skill for Claude Code

Data community

TradingAgents methodology as a portable, zero-dependency agent skill (A股/股票分析): multi-agent bull/bear debate + 5-tier rating gates + T+5 deferred reflection + A-share data playbook.

How to install Trading Agents Skill

This entry records only its repository, not the path inside it, so there is no exact command to give. Open yaorenliang/trading-agents-skill and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Trading Agents Skill does

TradingAgents methodology as a portable, zero-dependency agent skill (A股/股票分析): multi-agent bull/bear debate + 5-tier rating gates + T+5 deferred reflection + A-share data playbook. Apache-2.0.

Alternatives in Data

A Stock Data — A股全栈数据工具包 · 十一层架构 · 54端点 · 19数据源 · 零鉴权 Full-stack China A-share data toolkit for AI agents — 11 layers, 54 end 9.4k ★
HTML Video — Programmatic video for coding agents — HTML to video on your laptop 4.5k ★
Portaljs Add Resource — Add another file (resource) to an EXISTING dataset in a PortalJS portal — a data dictionary, methodology, or a 2.3k ★

README

trading-agents-sop

**A股特化的股票分析与交易决策 agent skill（A-share-first stock analysis & trading-decision skill）** — the TradingAgents methodology as a portable skill: multi-agent bull/bear debate, risk debate, centralized 5-tier rating gates, T+5 deferred reflection, and an A-share data-source playbook (akshare / 雪球 / 东财股吧 / 集思录). Pure Markdown, zero Python framework; any Agent-Skills-compatible harness (Claude Code, opencode, Codex, Gemini CLI, Cursor, DSH, ...) can execute it directly.

中文速览见文末。

Why this exists

[TradingAgents](https://github.com/TauricResearch/TradingAgents) (Tauric Research, Apache-2.0) is an excellent multi-agent LLM trading framework — but it is a Python/LangGraph pipeline: clone, install dependencies, configure API keys, run. If your daily driver is an AI coding agent, you do not want another environment to maintain.

This repo re-implements the **methodology** (not the code) as a skill, and adds a **discipline layer** the original does not expose in skill form:

	TradingAgents (framework)	trading-agents-sop (this skill)
Form	Python + LangGraph pipeline	Pure Markdown skill (Agent Skills format)
Install	clone + deps + API keys	copy one folder
Execution	full pipeline every run	tiered: quick path / full SOP / position review / reflection-only
Data	built-in US-centric connectors	data-source playbook: akshare / yfinance (free, tokenless), A-share sentiment sources (Xueqiu, Eastmoney guba, Jisilu)
Rating	5-tier via rating.py	centralized 5-tier gate + forced-decisiveness rule (references/rating.md)
Reflection	in-process memory log	T+5 deferred reflection with portable Markdown decision memory (readable across harnesses)
A-share specifics	limited	price-limit / T+1 / ST / unit / adjustment reminders

Install

OpenCode: copy to ~/.config/opencode/skills/trading-agents-sop/ (global) or .opencode/skills/ (per project).
Claude Code: copy this folder to `~/.claude