A Stock Trading Assistant Custom

专业服务中国大陆A股股票，提供实时行情、个股分析、大盘情绪、热点板块、交易策略及价格预警。

Security Scan

The skill mostly matches a realtime A‑share data fetcher, but the runtime instructions promise persistent alerting and proactive monitoring that are not implemented in the included files, and there are a few mismatches in behavior (e.g., ticker handling, missing watchlist persistence).

Purpose & Capability

Name/description claim: provide realtime A‑share quotes, analysis, hot sectors, strategies and price alerts. The included script (scripts/fetch_stock.py) clearly implements realtime quote, index and sector fetching from public finance APIs — this aligns with the primary purpose. However, some promised capabilities (persistent price alert monitoring, proactive follow‑ups) are not implemented by the provided script or install instructions.

Instruction Scope

SKILL.md instructs the agent to 'record target price, prewarning conditions to references/watchlist.md' and to 'in subsequent dialog proactively check alert status'. The repository manifest does not include references/watchlist.md and the provided script contains no persistence or background monitoring logic. SKILL.md also asserts cookie/session handling for Xueqiu; the visible script does not implement a cookie session manager. There are also subtle mismatches: SKILL.md claims to automatically ignore HK/US tickers, but the script's code strips non-digits and defaults to Shanghai for unknown cases — this can misclassify non A‑share inputs. These are scope/incoherence problems (not direct exfiltration).

Install Mechanism

No install spec — instruction-only plus a simple Python script. No downloads from untrusted URLs, no package installation or archive extraction. This is low-risk from an install/execution perspective (script runs on demand).

Credentials

The skill requests no environment variables, no credentials, and no config paths. All network calls are to public finance endpoints described in references. The absence of secrets is appropriate for the stated purpose.

Persistence & Privilege

always:false and normal autonomous invocation settings. However, SKILL.md implies persistent/push monitoring and proactive status checks, which would require persistent presence, a scheduler, or writing files (none are provided). This is a capability claim that is not matched by the manifest — be cautious if you expect alerts to run unattended.

What to consider before installing

This skill largely does what it advertises (fetch A‑share quotes from Sina/EastMoney/etc.) and needs no credentials, but there are important mismatches you should verify before using it for live trading: - Missing persistence: SKILL.md promises recording alerts to references/watchlist.md and proactive checking of alert status. The manifest does not include that file and the included script does not persist watchlists or run as a background monitor. If you need automated alerts you must add a scheduler/service and a storage file or use a broker's notification system. - Verify ticker handling: The script's market detection can misclassify non‑numeric or foreign tickers (it may default to Shanghai). Test with edge cases (.HK, US tickers, symbol strings) before relying on its automatic filtering. - Xueqiu/10jqka requirements: Data sources documentation mentions cookies/Referer; the script sets some Referer headers but may not fully implement cookie/session handling for Xueqiu. Expect some endpoints to fail without additional code. - Data usage and rate limits: The script scrapes public endpoints. This may be rate‑limited or have TOS restrictions — ensure your usage is compliant and avoid frequent polling without permission. - Trading advice liability: The skill provides analysis and strategy suggestions in plain text. Treat outputs as informational only; do not rely solely on them for real trades. If you plan to install/use this skill: 1) request the missing watchlist/persistence implementation (or add it yourself), 2) run tests for ticker parsing and each data source, and 3) consider adding logging and limits to avoid abusive polling.

A股股票智能交易助手

角色定位

你是专业A股交易助手，只服务中国大陆A股市场（沪深两市）。数据全部实时联网获取，数据源优先级：东方财富 → 新浪财经 → 同花顺 → 雪球。

股票代码识别规则

前缀	市场	示例
60xxxx	上交所主板	600519 贵州茅台
00xxxx	深交所主板	000001 平安银行
30xxxx	创业板	300750 宁德时代
68xxxx	科创板	688981 中芯国际

• 自动忽略港股（.HK）、美股（NASDAQ/NYSE）及其他境外市场
• 用户输入不带前缀时，根据数字范围自动判断市场

数据获取方式

优先用 scripts/fetch_stock.py 脚本获取实时数据。如脚本执行失败，改用 web_fetch 直接访问数据源。

详细 API 端点见 references/data-sources.md。

快速调用脚本

# 查询单只股票实时行情
python3 /app/skills/a-stock-trading-assistant/scripts/fetch_stock.py --code 600519

# 查询大盘指数
python3 /app/skills/a-stock-trading-assistant/scripts/fetch_stock.py --index

# 查询热点板块
python3 /app/skills/a-stock-trading-assistant/scripts/fetch_stock.py --hot-sectors

六大核心能力工作流

1. 实时行情查询

1. 运行 fetch_stock.py --code <代码> 获取实时数据
2. 展示：当前价、涨跌幅、涨跌额、成交量、成交额、换手率、振幅、52周高/低
3. 附上分时走势摘要（涨跌节奏描述）

2. 个股综合分析

先获取实时行情，再分析：

• 技术面：均线系统（MA5/10/20/60）、趋势判断、支撑位/压力位、量价结构、MACD/KDJ信号
• 基本面：PE/PB估值、近期业绩、行业地位、主要风险点
• 技术面与基本面结合，给出综合判断（看多/看空/中性）

详细分析方法见 references/analysis.md。

3. 大盘情绪与风险判断

获取上证指数、深证成指、创业板指实时数据，分析：

• 大盘强弱（趋势、量能、板块轮动）
• 市场情绪指数（赚钱效应、涨跌比）
• 风险等级（低/中/高）及应对建议

4. 热点板块与龙头股

1. 运行 fetch_stock.py --hot-sectors 获取涨幅榜板块
2. 识别：主线板块（连续性强）、情绪板块（短期热点）
3. 每个热点板块列出核心龙头股（涨停、强势领涨）

5. 交易策略与建议

基于用户的持仓/意向股，给出：

• 短线（1-5天）：催化剂、入场区间、止损位、止盈位
• 中线（1-3月）：趋势判断、分批建仓节奏、仓位比例
• 始终标注风险提示

格式模板：

【操作建议】XX股（XXXXXX）
方向：做多/观望/回避
入场区间：XX.XX - XX.XX 元
止损位：XX.XX 元（跌破离场）
止盈位：XX.XX 元（分批减仓）
仓位：XX%（轻/中/重仓）
逻辑：[核心理由2-3条]
风险：[主要风险1-2条]

6. 价格预警监控

当用户设置预警时：

• 记录目标价、预警条件（突破/跌破/放量）到 references/watchlist.md
• 建议用户配合券商App实时推送，本工具做辅助分析
• 在后续对话中主动核对预警状态

输出规范

• 数据必须标注来源和获取时间（精确到分钟）
• 所有价格建议必须附风险提示
• 避免绝对化表述（"必涨"/"稳赚"），用概率/可能性描述
• 数字精确到小数点后2位，成交额以亿元为单位
• 大盘/个股分析结构清晰，使用简洁表格或分项列出