futu-unusual-capital-flows
Unusual Capital Flows (资金异动): detects capital-flow unusual signals for a user-mentioned stock with `get_financial_unusual`, including 资金分布、买卖经纪商、 资金流向、卖空数量、卖空比例 and their unusual changes. Use when the user asks about 资金动向、资金异动、净流入、净流出、主力行为、大单小单、谁在买谁在卖、买卖经纪商、 卖空异动、卖空/做空/沽空、主力是否在进场或离场,or gives a broad 异动 request such as “英伟达异动”、“NVDA 异动”、“腾讯有没有异动/异常” without narrowing the dimension. For broad 异动 requests, use together with `futu-unusual-technical-signal-alerts` and `futu-unusual-derivatives-activity`. Before calling the CLI, you must first normalize the user-mentioned stock name, Chinese company name, English company name, or ticker into a standard symbol such as `US.TSLA` or `HK.00700`.
How do I install this agent skill?
npx skills add https://github.com/futunnopen/futu-agent-hub --skill futu-unusual-capital-flowsIs this agent skill safe to install?
- Gen Agent Trust Hubpass
The skill provides a specialized interface for checking unusual financial capital flows using a provided command-line tool. It follows secure practices by requiring stock symbol normalization, avoiding hardcoded secrets, and using a local CLI with OAuth-managed authentication.
- Socketpass
No alerts
- Snykpass
Risk: LOW · No issues
What does this agent skill do?
资金异动 · Unusual Capital Flows
Detects unusual capital-flow signals for a specific stock and formats the result as a structured Unusual Capital Flows summary.
This skill is for unusual-move detection rather than a regular capital-flow overview. If the data has no qualifying signal, return No unusual activity or the interface's no-data result, and do not add extra market commentary.
If the user only says a broad request such as 英伟达异动, NVDA 异动, 腾讯有没有异常 and does not specify a dimension, treat it as a bundled 异动 request. In that case, use this skill together with futu-unusual-technical-signal-alerts and futu-unusual-derivatives-activity.
Prerequisite: Futu CLI
This skill calls futu unusual capital-flows (OAuth REST via the official CLI). Do not use OpenD, futu-api, or removed futuapi scripts.
Before the workflow, follow CLI runtime. If futu unusual is not recognized, say Unusual skills need a newer CLI — see repo docs/migration-gaps-and-rest-api.md.
Workflow
1. Parse User Input
Extract the following from the user's request:
stock_target: stock code, Chinese stock name, English company name, or ticker explicitly mentioned by the usertime_range: default7; if the user says最近 3 天/过去两周/last 5 days, convert it to a natural-day integeranalysis_dimensions: optional; only extract when the user clearly asks about one or more specific financial dimensionslanguage_id: infer from the user's language
If the target stock is missing, ask a follow-up question instead of guessing.
2. Normalize the Stock Target into a Standard Symbol
Before calling the CLI, convert the user-mentioned stock target into a standard symbol such as US.TSLA, HK.00700, SH.600519, or SZ.000001.
Normalization rules:
- If the user already gives a fully qualified symbol like
US.TSLAorHK.00700, use it directly. - If the user gives a Chinese company name, English company name, or common ticker, map it to the matching market-prefixed symbol.
- If the symbol is ambiguous, ask a follow-up question instead of guessing.
Common mappings:
| User mention | Standard symbol |
|---|---|
| 腾讯 | HK.00700 |
| 阿里巴巴、阿里 | HK.09988 |
| 苹果、Apple | US.AAPL |
| 特斯拉、Tesla | US.TSLA |
| 英伟达、NVIDIA | US.NVDA |
| 微软、Microsoft | US.MSFT |
| 谷歌、Google、Alphabet | US.GOOG |
| 亚马逊、Amazon | US.AMZN |
| Meta、脸书、Facebook | US.META |
| 台积电、TSM | US.TSM |
| 贵州茅台、茅台 | SH.600519 |
| 宁德时代 | SZ.300750 |
Ticker inference guidance:
TSLA,AAPL,NVDA,MSFT,GOOG,METAusually mean US stocks, so normalize toUS.TSLA,US.AAPL,US.NVDA,US.MSFT,US.GOOG,US.META.00700by itself is ambiguous in plain text; when the context clearly refers to Tencent, normalize toHK.00700.- If the same company has both HK and US listings and the user does not specify the market, ask a clarifying question.
3. Map Financial Intent to analysis_dimensions
Prefer these canonical analysis_dimensions values:
funds_distribution: Capital distributionfunds_broker: Buy/sell brokersfunds_flow: Capital flowshort_sell_number: Short sell volumeshort_sell_ratio: Short sell ratioshort_sell_number_and_ratio: Short sell volume and ratio anomalous simultaneously
Selection guidance:
- If the user asks about
谁在买/谁在卖/经纪商, includefunds_broker. - If the user asks about
大单小单/资金分歧/资金分布, includefunds_distribution. - If the user asks about
主力连续流入流出/资金流向, includefunds_flow. - If the user asks about
卖空数量, includeshort_sell_number. - If the user asks about
卖空比例, includeshort_sell_ratio. - If the user asks about
卖空数量和比例是否同时异常, includeshort_sell_number_and_ratio. - If the user asks a broad capital-flow question and does not narrow scope, omit
analysis_dimensionsand use the full interface.
4. Infer language_id
Use the following language mapping:
0: Simplified Chinese1: Traditional Chinese2: English4: Thai5: Japanese
Default strategy:
- Chinese user ->
0 - Traditional Chinese request ->
1 - English user ->
2 - If unclear, default to
0
5. Call the CLI
futu unusual capital-flows <STANDARD_SYMBOL> --time-range <DAYS> [--analysis-dimensions <DIM> ...] [--language-id <ID>] --format json --source skill
Examples:
futu unusual capital-flows US.NVDA --time-range 7 --format json --source skill
futu unusual capital-flows HK.00700 --time-range 7 --analysis-dimensions funds_distribution funds_broker --format json --source skill
futu unusual capital-flows US.AAPL --time-range 5 --analysis-dimensions funds_flow --format json --source skill
6. Check the Result
- If the CLI exits successfully, use the JSON to build a structured capital-flow anomaly summary.
- If the CLI returns an error, surface the error message and do not fabricate results.
- If a requested class has no anomaly in the window, explicitly say
No anomaly. - If the downstream service reports a permission problem or no accessible data, clearly state that the stock or account lacks the required permission.
Parameters
Required
stock_symbol: standard market-prefixed symbol, such asUS.TSLA,HK.00700,SH.600519,SZ.000001
Optional
time_range: natural day window, default7analysis_dimensions: list of specific financial anomaly dimensions to inspect; omit for full scanlanguage_id: output language, default0
Output Rules
Present the output by anomaly class. The three high-level classes are:
Capital Distribution & Buy/Sell BrokersCapital FlowShort Selling
Formatting rules:
- Always display
Time Rangeas an absolute date range in the formatYYYY.M.D - YYYY.M.D. Calculate the start date from the current date minustime_rangedays (e.g., if today is 2026.4.24 and time_range is 7, writeTime Range: 2026.4.18 - 2026.4.24). - If multiple abnormal dates appear within the window for one class, list them all.
- If a class has no anomaly in the window, write
No anomaly. - Preserve dates, direction, amount, ratio, broker names, and interpretation from the tool output.
- Do not merge different anomaly classes into one sentence.
- Do not invent thresholds, rankings, or causal explanations beyond the returned content.
- If the user only asked about a subset of financial dimensions, keep only the relevant classes.
Preferred Response Template
Time Range: {YYYY.M.D - YYYY.M.D}
- Capital Distribution & Buy/Sell Brokers: {anomaly description or "No anomaly"}
- Capital Flow: {anomaly description or "No anomaly"}
- Short Selling: {anomaly description or "No anomaly"}
When the user only asks for a subset, keep only the relevant classes.
Behavior Rules
- Always normalize the user-mentioned stock target into a standard symbol before calling the CLI.
- If the target stock is missing, ask a follow-up question.
- If the market is ambiguous, ask a follow-up question instead of guessing.
- If the user asks a broad capital-flow question, omit
analysis_dimensionsand use the full interface. - Do not output raw JSON by default when talking to the user, even if the CLI returns JSON.
- Do not interpret the result as investment advice or trading guidance.
- If the downstream service reports a permission issue, state that clearly and do not fabricate analysis.
Example User Requests
Broad capital anomaly check
- "特斯拉最近 7 天有没有资金面异动?"
- "看看腾讯最近主力有没有异常动作"
- "帮我查一下 Apple 最近资金面有没有异常"
Mapped request:
{
"stock_symbol": "US.TSLA",
"time_range": 7,
"language_id": 0
}
Broker and capital-distribution only
- "腾讯最近谁在买谁在卖?"
- "00700 最近大单小单有没有分歧?"
Mapped request:
{
"stock_symbol": "HK.00700",
"time_range": 7,
"analysis_dimensions": ["funds_distribution", "funds_broker"],
"language_id": 0
}
Funds flow only
- "苹果最近主力资金有连续流入流出吗?"
Mapped request:
{
"stock_symbol": "US.AAPL",
"time_range": 5,
"analysis_dimensions": ["funds_flow"],
"language_id": 0
}
Short sell only
- "TSLA 最近卖空比例有没有异常?"
Mapped request:
{
"stock_symbol": "US.TSLA",
"time_range": 7,
"analysis_dimensions": ["short_sell_ratio"],
"language_id": 0
}
Example Interpretation Style
Time Range: 2026.4.2 - 2026.4.9
- Capital Distribution & Buy/Sell Brokers: On 4.3, extra-large orders had a net inflow of 113 million yuan with inflow/outflow magnitudes differing by more than 2x, while small orders had a net outflow of 178 million yuan in the opposite direction, indicating a divergence between large and small capital. From the broker side, the top two buyers were China Investment (Shanghai-HK Connect) and Futu Securities, while the top two sellers were UBS and Barclays Asia.
- Capital Flow: On 4.4, main capital had net outflows for 4 consecutive days, and that day's net outflow was 120% higher than the previous 3-day average, indicating main capital is accelerating its exit.
- Short Selling: On 4.5, short sell volume and short sell ratio were both anomalous: short sell volume rose day-over-day and the short sell ratio rose in parallel, both above the recent one-month average, reflecting a notably bearish outlook.
How can the creator link this skill?
Add the canonical catalog link to the repository README so users can inspect current installs and available audits. The publishing guide covers the complete discovery path.
<a href="https://skillzs.dev/skills/futunnopen/futu-agent-hub/futu-unusual-capital-flows">View futu-unusual-capital-flows on skillZs</a>