Guide · Agent skill
How to use the US ETF Flow Analysis agent skill
Updated · written with AI from the project's own documentation and checked against our catalogue data
The us-etf-flow skill gives an AI agent a framework for reading money moving into and out of US exchange-traded funds, and using that as a clue about where large investors are positioning. It is written for investors, analysts, students and developers who want a structured way to think about ETF flows around sector rotation. After this guide you will know what the skill does, how to install it, how to trigger it with plain requests, and what it does and does not need to work.
US ETF Flow AnalysisHKUDS/Vibe-Trading · MIT Download the skill Quick start
- 1Download the ZIP. Get the us-etf-flow ZIP from this page and unzip it so you have the SKILL.md file.
- 2Place the folder. Move the unzipped folder into ~/.claude/skills/ for personal use or into a project's .claude/skills/ folder.
- 3Or upload in the apps. In the Claude apps, go to Settings → Capabilities → Skills and upload the skill.
- 4Prepare your flow data. Gather daily ETF flow and AUM data yourself, since the skill names no data source.
- 5Ask a plain question. Describe the sector, ETF or rotation question you want answered and mention flows.
- 6Check the thresholds. Ask the agent which hard-coded thresholds it used and confirm they suit your data.
What is an agent skill, and what does this one make the agent do?
An agent skill is a folder containing a SKILL.md file with instructions that an AI agent reads when a task matches the skill's description. This skill, named us-etf-flow, contains analytical guidance about US ETF flows, sector breadth and style factor flows.
Once installed, the agent can explain how ETF creation and redemption reflects demand, group flows by broad market, sector, style, bond and thematic ETFs, and build simple breadth and cyclical-versus-defensive indicators from flow data. The skill includes example Python logic for these calculations. It does not fetch live market data itself.
Install the us-etf-flow skill
Download the ZIP file from this page and unzip it. The skill's files are listed as SKILL.md. Then choose one of two locations. For a personal install, place the unzipped folder in ~/.claude/skills/. For a single project, place it in that project's .claude/skills/ folder.
In the Claude apps you can instead upload the skill under Settings → Capabilities → Skills.
- Personal install: ~/.claude/skills/
- Project install: .claude/skills/ inside your project
- Claude apps: Settings → Capabilities → Skills, then upload
Core ideas the skill teaches
Creation happens when authorized participants deliver baskets of underlying securities to the ETF issuer and receive new ETF shares, which usually happens when the ETF trades at a premium to NAV. Redemption is the reverse and usually happens at a discount to NAV. The skill treats sustained large inflows as institutional demand and sustained large outflows as institutional liquidation.
The skill also stresses that ETF price movement is not the same as ETF flow. Price can rise on low volume, while flows require actual capital commitment. Flows are a quantity signal, not a price signal.
- Inflow (creation): ETF trades at a premium to NAV
- Outflow (redemption): ETF trades at a discount to NAV
- Price move is not flow; flow measures committed capital
How to trigger the skill with natural requests
You do not need special syntax. Describe the question in plain language and mention flows, sectors, or ETFs. The agent should match the request to the skill's description, which covers ETF fund flows, sector rotation breadth, style factor flows, and thematic momentum.
Supply your own flow and AUM data in the request or in a file, because the skill does not name a data source.
- Which sectors are seeing ETF inflows versus outflows this month, and what does the breadth say?
- Is money rotating from cyclical ETFs like XLY and XLI into defensive ones like XLP and XLU? Use my flow table.
- Calculate the cyclical versus defensive ratio from these daily flows for the sector ETFs.
- Using the flow data I pasted, is there a strong_inflow or mild_outflow signal for SMH over 20 days?
Examples of what the agent will produce
Given sector flow figures, the skill's breadth formula counts the share of sectors with positive flows. Its interpretation rules are: above 0.7 suggests broad inflows, 0.4 to 0.7 suggests selective rotation, and below 0.4 suggests broad outflows. The cyclical versus defensive ratio compares summed flows of cyclical sectors (XLK, XLY, XLI, XLF, XLB) with defensive sectors (XLV, XLP, XLU, XLRE).
The skill also covers broad market flow combinations, such as SPY and IWM both receiving inflows, and a single-ETF signal function that compares cumulative flow as a percentage of AUM and recent momentum.
- Breadth reading from the sector flow dictionary in the file
- Cyclical vs defensive ratio with its stated thresholds
- Single-ETF signal labels: strong_inflow, mild_inflow, neutral, mild_outflow, strong_outflow
positive_sectors = sum(1 for v in sector_flows.values() if v > 0)
breadth = positive_sectors / len(sector_flows)What you need before using it
The skill requires pre-collected ETF daily flow and AUM data supplied by you. No data source is specified in the skill. Python is needed only if you ask the agent to run the example formulas.
The skill's data-access section shows a yfinance example for price, volume and some info fields such as total assets. It notes that sector weights are not directly available through yfinance. Flow data sources listed in the skill include ETF.com (free web access, T+1 latency) and Bloomberg Terminal (paid, real-time).
- Daily ETF flow and AUM data you provide
- Python, if you want the example formulas run
- Optional: yfinance for price and volume
- The skill's flow source table lists ETF.com and Bloomberg Terminal
Safety notes: scripts and network
This skill does not run scripts and does not use the network, according to its listing. It is a set of instructions and example code. Treat the example Python as illustrations you should review and test before relying on them.
The thresholds are hard-coded and simplified. The skill's own text describes a rule of thumb that 3-standard-deviation flow events in sector ETFs tend to mean-revert within 2-4 weeks. Verify any such claim against your own data before acting on it.
- Runs scripts: no
- Uses network: no
- Hard-coded thresholds: review them before use
- Flows alone do not predict prices
Tips for getting useful answers
Give the agent clean, dated flow and AUM data for the same ETFs and window. Ask it to state which threshold it applied. Keep momentum and contrarian readings separate, since the skill says momentum tends to work for broad market ETFs in trending markets while contrarian fading suits sector and thematic ETFs at extremes.
Use the thematic guidance as a lens. The skill says sustained four-week-plus inflows suggest conviction rather than hot money, and that sudden large outflows from a trending theme can signal crowded-trade unwind risk.
- Provide consistent data for the same tickers and dates
- Ask the agent to show the threshold it used
- Compare flow readings with price before drawing conclusions
What people use it for
- Reading whether money is rotating from cyclical sector ETFs into defensive ones
- Measuring sector breadth from a table of daily sector ETF flows
- Learning how creation and redemption connect to premiums and discounts to NAV
- Drafting starting formulas for a flow-monitoring dashboard
- Classifying a single ETF's recent flows as inflow, neutral or outflow
Frequently asked questions
Is the us-etf-flow skill free?
The skill is published under the MIT licence, as listed on this page. The flow data you need is separate: the skill names no data source, and some sources it mentions are paid.
What do I need to run the us-etf-flow skill?
You need pre-collected daily ETF flow and AUM data supplied by you, and Python if you want the example formulas run. The skill does not fetch data or run scripts on its own.
How do I install the us-etf-flow agent skill?
Download the ZIP from this page and unzip it into ~/.claude/skills/ or a project's .claude/skills/ folder. In the Claude apps you can upload it under Settings → Capabilities → Skills.
Does the skill give live ETF flow data?
No. The skill is analytical guidance with example Python logic and example figures. You must supply the flow data, and the skill's listed sources include ETF.com and Bloomberg Terminal.
Can I use ETF flows to predict stock prices?
The skill itself says flows alone do not predict prices and that flows measure quantity, not price. Use them as one input alongside independent validation.
How does this compare with other flow tools or data terminals?
The skill is a reasoning framework that runs inside an agent, not a data feed or terminal. If you need live data, pair it with a source such as a paid terminal or a web data provider, or choose a dedicated flow dashboard.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
Use this skill if you want a clear framework for reading ETF flows around sector rotation, or starting formulas for a dashboard, and you can supply your own data. Pick something else if you need live market data, a tested trading system, or validated predictive signals, because the rules here are simplified illustrations with hard-coded thresholds.
Download the skill