Claude-skill-registry diagnose

install

source · Clone the upstream repo

git clone https://github.com/majiayu000/claude-skill-registry

Claude Code · Install into ~/.claude/skills/

T=$(mktemp -d) && git clone --depth=1 https://github.com/majiayu000/claude-skill-registry "$T" && mkdir -p ~/.claude/skills && cp -r "$T/skills/data/diagnose" ~/.claude/skills/majiayu000-claude-skill-registry-diagnose && rm -rf "$T"

manifest: skills/data/diagnose/SKILL.md

Trading System Diagnostics

Purpose

Use this skill when:

No trading signals are being generated
Need to verify AI analysis is working
Validating technical indicator calculations
Debugging market data issues

Diagnostic Commands

Full Diagnostic (Default)

cd /home/linuxuser/nautilus_AItrader
source venv/bin/activate
python3 scripts/diagnose.py

Quick Diagnostic (Skip AI calls)

cd /home/linuxuser/nautilus_AItrader
source venv/bin/activate
python3 scripts/diagnose.py --quick

With Update and Restart

python3 scripts/diagnose.py --update --restart

Expected Output

Normal Operation Signs

✅ Configuration loaded successfully
✅ Market data fetched successfully
✅ TechnicalIndicatorManager initialized
✅ Technical data retrieved
✅ Sentiment data retrieved
✅ MultiAgent 层级决策成功
   🐂 Bull Agent 分析中...
   🐻 Bear Agent 分析中...
   ⚖️ Judge Agent 判断中...
   🛡️ Risk Manager 评估中...
🎯 Judge 最终决策: BUY/SELL/HOLD

Key Checkpoints

Check	Normal Value	Abnormal Handling
RSI	0-100	Out of range = data error
MACD	Any value	NaN = insufficient data
Judge Signal	BUY/SELL/HOLD	ERROR = API failure
Winning Side	BULL/BEAR/TIE	显示辩论胜方

信号决策流程 (层级决策架构)

v6.0 更新: 采用 TradingAgents 层级决策架构，Judge 决策即最终决策

决策流程:
Phase 1: Bull/Bear Debate (辩论)
  └→ 🐂 Bull Agent: 寻找做多理由
  └→ 🐻 Bear Agent: 寻找做空理由

Phase 2: Judge (Portfolio Manager) Decision
  └→ ⚖️ 评估辩论结果，做出最终决策

Phase 3: Risk Evaluation
  └→ 🛡️ 确定仓位大小和止损止盈

注意: 以下配置已标记为 LEGACY，不再生效:

skip_on_divergence: true      # [LEGACY] 不再使用
use_confidence_fusion: true   # [LEGACY] 不再使用

Common Issues

1. No Trading Signals

Possible Causes:

Judge returns HOLD (Bull/Bear辩论无明显胜者)
Confidence below min_confidence_to_trade
Risk Manager 认为风险过高

Check Command:

python3 scripts/diagnose_realtime.py 2>&1 | grep -E "(Judge|Final Signal|Confidence|Winning Side)"

2. DeepSeek API Failure

Check:

grep "DEEPSEEK_API_KEY" ~/.env.aitrader

3. Abnormal Technical Indicators

Check:

python3 scripts/diagnose.py 2>&1 | grep -E "(RSI|MACD|SMA)"

Key Files

File	Purpose
`scripts/diagnose.py`	Main diagnostic script
`scripts/diagnose_realtime.py`	Real-time API diagnostic
`scripts/smart_commit_analyzer.py`	Regression detection (auto-evolving rules)
`strategy/deepseek_strategy.py`	Main strategy logic
`configs/base.yaml`	Base configuration (all parameters)
`configs/production.yaml`	Production environment overrides

回归检测 (修改代码后必须运行)

# 智能回归检测 (规则自动从 git 历史生成)
python3 scripts/smart_commit_analyzer.py

# 预期结果: ✅ 所有规则验证通过