菜单
Use this skill when the user asks how much to invest, wants to plan or enter a trade, mentions position sizing, asks about their trading discipline or losses, or reviews their trade journal. Also use before endorsing any specific entry the user proposes.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Use this skill when the user asks to backtest a strategy, evaluate historical performance, compare strategies, interpret Sharpe ratio or drawdown, validate a strategy's statistical reliability, or run walk-forward analysis. Also use when recommending any systematic trading strategy that should be validated against historical data before being proposed.
Use this skill when the user asks about pair trading, statistical arbitrage, cointegrated pairs, spread trading, hedge ratios, or market-neutral strategies.
Use this skill when the user asks about position sizing, stop-loss levels, portfolio risk, VaR, drawdown limits, risk/reward ratios, or when you are about to recommend an entry point for any trade. Always apply this skill before stating a specific price target, stop, or position size.
Use this skill when the user asks which strategy to use, how to build a trading system, how to select between trend-following and mean-reversion, what works in the current market regime, or when designing an entry/exit/stop framework for any equity or ETF. Also use when the user describes market conditions and wants a structured approach.
Use this skill when the user asks how a stock behaves around earnings, wants an earnings calendar, is planning a trade into or after an earnings report, or mentions post-earnings drift (PEAD).
Use this skill when the user asks for a report, brief, summary document, or export — daily market brief, sector report, stock deep-dive, portfolio review, or screening results — especially when they want it saved to a file or in HTML.
| name | exposure-discipline |
| description | Use this skill when the user asks how much to invest, wants to plan or enter a trade, mentions position sizing, asks about their trading discipline or losses, or reviews their trade journal. Also use before endorsing any specific entry the user proposes. |
| license | MIT |
| metadata | {"author":"quantagent","version":"1.0"} |
| allowed-tools | synthesize_conviction_tool, check_risk_circuit_breaker, check_trade_discipline, journal_log_trade, journal_update_status, journal_close_trade, journal_open_trades, journal_history, journal_stats, detect_market_regime |
Process beats prediction. This skill enforces a layered discipline stack before any entry: market context → account guardrails → per-trade checklist. Every recommendation ends with an explicit exposure or go/no-go statement, never just analysis.
synthesize_conviction_tool (or
detect_market_regime). The conviction score maps to an equity
exposure band; new longs in a defensive/risk-off stance need an
explicit justification and reduced size.check_risk_circuit_breaker. Rules: daily loss
2%, weekly >5%, monthly >8% (halted), or 2 consecutive losers (24h cooldown). If tripped, the answer to "should I take this trade?" is no, and here's when the cooldown lifts — regardless of how good the setup looks. That is the point of the breaker.
check_trade_discipline on the journaled idea.
Blocks without a written thesis, entry plan, or stop, and in
reduce-only regimes.journal_log_trade) before entry:
symbol, falsifiable thesis, entry plan, target, stop.journal_close_trade and an honest postmortem note. MAE/MFE
are captured automatically — use them in reviews: consistently large
MAE means entries are early or stops too wide; large unrealized MFE
given back means exits are too slow.journal_stats): expectancy and profit factor matter more
than win rate. Flag max consecutive losses vs the cooldown setting.