菜单
Create, update, inspect, and validate persistent watchlists stored as JSON files. Triggers on "自选股", "watchlist", "监控清单", "add to my list", "把XX加入监控", "建一个组合清单".
默认使用会先检查来源的 Prompt；你也可以切换为直接命令，或下载本地副本。
决定是否安装前，请先阅读 SKILL.md，以及 SkillsMP 当前展示的配套文件。
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。
直接命令不会经过审查 Prompt；运行前请先检查来源。
npx skills add https://github.com/zai-org/zcode-plugins --skill watchlist命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地？可下载 SkillsMP 当前能够提供的文件。
A mapping between two charts of accounts — multi-book consolidation, a system migration, or a presentation change — with balance conservation proved rather than assumed, one-to-many and many-to-one relations surfaced for confirmation instead of resolved silently, and every 口径 change named. Triggers on "科目映射", "科目对照", "两个账套怎么合", "新旧科目转换", "科目表迁移", "重分类", "报表口径调整", "合并口径", "chart of accounts mapping", "account reclassification".
Audit a spreadsheet for formula accuracy, errors, and common mistakes. Scopes to a selected range, a single sheet, or the entire model (including financial-model integrity checks like BS balance, cash tie-out, and logic sanity). Triggers on "audit this sheet", "check my formulas", "find formula errors", "QA this spreadsheet", "sanity check this", "debug model", "model check", "model won't balance", "something's off in my model", "model review".
Statutory-basis financial statements and note working papers built from a trial balance — accounts mapped to the caption format the stated basis prescribes, the three statements with every tie shown as a live formula, note schedules traceable to the accounts behind them, and every accounting judgement registered for the finance lead rather than decided. Delivered as a draft pending review, unaudited. Triggers on "编制财务报表", "出一版财务报表", "法定报表", "准则报表", "资产负债表利润表现金流量表", "合并报表编制", "年报报表", "报表底稿", "附注底稿", "financial statements", "statement preparation".
正在显示 SKILL.md
| name | watchlist |
| description | Create, update, inspect, and validate persistent watchlists stored as JSON files. Triggers on "自选股", "watchlist", "监控清单", "add to my list", "把XX加入监控", "建一个组合清单". |
Watchlists live in watchlists/<list-name>.json in the working directory. One list per file.
Create / add names / remove names / show / rename. If the user names a list that doesn't exist and the operation is "add", create it and say so.
.SH / .SZ, 港股 .HK, 美股 .O / .N — or company names/aliases.hexin-stock.search_stocks (query = 简称/名称); if a name is ambiguous (multiple listings, A/H dual listing), ask which one before writing.Schema:
{
"name": "<list-name>",
"updated": "<YYYY-MM-DD, today>",
"positions": [
{"windcode": "[证券代码]", "label": "[证券简称]", "weight": null, "cost": null, "notes": ""}
]
}
weight / cost / notes on names you are not touching.weight is a decimal fraction (0.05 = 5%); cost is per-share in listing currency. Only set them when the user supplies them.Show the resulting list as a short table (windcode, label, weight if any) and cite the JSON deliverable exactly once with ::zcode-file-citation{path="..." purpose="output"} inline in prose. Do not add a separate raw path, Markdown link, or trailing citation list. If the list exceeds ~30 names, warn that recap and event scans will be slower and suggest splitting.
The written JSON is the deliverable, so deliver it. This skill's product is a
saved file, not a message about one: put watchlists/<list-name>.json where the
user collects output, the same as any other artifact, rather than leaving it only
in the working directory. A run that resolved every code correctly and handed back
nothing looks identical to a run that did no work — and the list is what the user
will point close-recap, intraday-watch, and event-monitor at next.
Read the file back after writing and show the parsed contents, so the codes the user confirms are the ones actually on disk rather than the ones you intended to write.