# Agenticx Query Data Source > Use when the user asks about verifiable quantitative facts (stock prices, financial indicators, macro data, company registry, academic metrics, legal statutes) that must come from a live data source rather than training memory. - Skill: `demondamon/agenticx-query-data-source` (Agent Skill) - Install (CLI): `npx skillmds@latest add demondamon/agenticx-query-data-source` - Raw SKILL.md: https://api.skillmd.com/api/skills/demondamon/agenticx-query-data-source/raw - Safety review: pending - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: Coding & Dev Tools - Author: DemonDamon (https://skillmd.com/u/demondamon) - Updated: 2026-09-22 - Page: https://skillmd.com/skills/demondamon/agenticx-query-data-source --- # Query Data Source ## When to use - Any question about **stock price**, **financial indicators**, **macro economic data**, **company registry**, **academic citation counts**, or **legal statute text** that has a real, checkable current value. - **Do NOT** answer from training memory for these categories — training data is stale and the user can verify against a real source. - Call `list_data_sources` first when unsure which plugin or API fits (optional `domain`: `finance` / `macro` / `academic` / `enterprise` / `legal`). ## How to call 1. `list_data_sources(domain="finance")` — discover enabled plugins and `api_name` values. 2. `query_data_source(data_source_name=..., api_name=..., params={...})` — fetch structured JSON. - **Default chart window: `days: 60` (≈3 个月交易日).** A candlestick chart with only 5–10 bars looks sparse/empty in the chat widget, so this is the default for any "走势/行情/表现如何" style question — including casual phrasing like "最近一周走势" (interpret as "show me how it's been doing lately", not literally 5 trading days). - Only request a shorter window when the user gives an **unambiguous analytical** ask for an exact short range (e.g. "对比昨天和前天的收盘价" or "最近3个交易日的具体数字"). Map common asks: `1个月→days:20`, `3个月→days:60`（默认）, `6个月→days:120`, `1年→days:250`. 3. For **time-series** results (price history, macro trend), **must** follow with `show_widget`: - Prefer structured JSON (Desktop renders via ECharts): - **The top-level `"type": "stock_chart"` field is MANDATORY.** Omitting it makes the Desktop client treat the payload as a generic HTML widget and render the raw JSON text instead of a chart — always double-check this field is present before calling `show_widget`. **Single stock:** ```json { "type": "stock_chart", "title": "火炬电子 603678", "chart_type": "candlestick", "data_source_label": "获取数据 | AkShare（免费行情）", "points": [ {"date": "2026-07-01", "open": 80, "high": 90, "low": 78, "close": 85, "volume": 120000} ], "attribution": "数据来源：AkShare" } ``` **Multiple focused stocks (Kimi-style tabs — user can switch 火炬电子 / 祥鑫科技 / 天齐锂业):** ```json { "type": "stock_chart", "data_source_label": "获取数据 | AkShare（免费行情）", "attribution": "数据来源：AkShare", "watchlist": [ { "symbol": "603678.SH", "name": "火炬电子", "chart_type": "candlestick", "points": [{"date": "2026-07-03", "open": 81.26, "high": 89.4, "low": 77.93, "close": 85.48, "volume": 42942215}] }, { "symbol": "002965.SZ", "name": "祥鑫科技", "chart_type": "candlestick", "points": [...] }, { "symbol": "002466.SZ", "name": "天齐锂业", "chart_type": "candlestick", "points": [...] } ] } ``` When the user mentions several tickers they are tracking, fetch each via `query_data_source` and pass **one** `show_widget` with a `watchlist` array — do not render three separate widgets. > **CRITICAL — never hand-roll HTML/ECharts for stock charts.** Always pass the structured > `stock_chart` JSON (single `points` or `watchlist`). The Desktop client has a dedicated, > theme-aware `StockChartWidget` (dark/light adaptive tabs, 红涨绿跌, quote header). Hand-written > `
`+ECharts `