Guide · Agent skill
How to use the minute-analysis agent skill (VWAP, TWAP)
Updated · written with AI from the project's own documentation and checked against our catalogue data
The minute-analysis skill is a set of instructions and a Python example that lets an AI agent pull minute-by-minute price candles from OKX, Tushare or yfinance and calculate intraday measures such as VWAP, TWAP and volume distribution. This guide is written for readers who have not used agent skills before. By the end you will know what the skill does, how to install it, how to ask for it in plain language, and which limits to check before relying on its output.
Minute-Level Market AnalysisHKUDS/Vibe-Trading · MIT Download the skill Quick start
- 1Download the ZIP. Download the minute-analysis ZIP file from this page.
- 2Unzip into a skills folder. Extract it into ~/.claude/skills/ for personal use, or into a project's .claude/skills/ folder.
- 3Or upload in the Claude apps. In the Claude apps, go to Settings, then Capabilities, then Skills, and upload the ZIP.
- 4Install Python dependencies. Run pip install pandas numpy requests so the example code can run.
- 5Ask in plain language. Name the instrument, interval and time window, for example BTC-USDT on 5m candles for the last week.
- 6Check the output. Confirm the time range and row count, then review the VWAP, TWAP or volume figures.
What is an agent skill and what does minute-analysis do?
An agent skill is a folder containing a SKILL.md file and any helper scripts. When you install it, an AI agent that supports skills can read the instructions and use them when a request matches the skill's description.
Once minute-analysis is installed, the agent can retrieve minute candlesticks (open, high, low, close and volume at 1m, 5m, 15m, 30m, 1H or 4H intervals) and compute intraday indicators from them. The source describes VWAP (volume-weighted average price), TWAP (time-weighted average price) and volume distribution.
- OKX: cryptocurrency pairs such as BTC-USDT, trading 7 days a week, no API key needed for public candle data
- Tushare: China A-shares, requires a Tushare score of at least 2000
- yfinance: Hong Kong and US equities, free and no key required
How do you install the minute-analysis skill?
Download the ZIP file for the skill from this page. Unzip it so that the minute-analysis folder sits in your personal skills directory, ~/.claude/skills/, or in a project's .claude/skills/ folder if you want it available only for that project.
Alternatively, in the Claude apps go to Settings, then Capabilities, then Skills, and upload the ZIP there.
Install the Python dependencies the skill uses. The skill needs pandas, numpy and requests.
pip install pandas numpy requestsHow do you trigger the skill with a natural request?
You do not need a special command. Describe the data and the measure you want in ordinary language, and name the market if it helps the agent choose a data source. The skill's description mentions minute-level analysis, OKX, Tushare and yfinance, so requests that mention these are most likely to match.
Requests that work well state the instrument, the interval and the time window. Be realistic about the window: the skill's own guidance suggests no more than 7 days for 1m data, 30 days for 5m and 1 year for 1H.
- Calculate VWAP and TWAP for BTC-USDT on 5-minute candles over the last week from OKX.
- Show the volume distribution by hour for BTC-USDT on 1H candles this month.
- Fetch 1-minute candles for AAPL with yfinance and compute intraday VWAP.
- Compare intraday volume patterns between two OKX trading pairs over the past 5 days.
How does the skill fetch OKX minute candles?
The skill's OKX example calls the public market candles endpoint with an instrument ID (instId), a bar size (bar) and a limit. OKX returns at most 300 rows per request, so the loader paginates automatically. The timestamps are Unix milliseconds and are converted to datetimes.
Note that 1m datasets are still very large even with pagination, so keep the time range short.
resp = requests.get("https://www.okx.com/api/v5/market/candles", params={"instId": "BTC-USDT", "bar": "1m", "limit": "300"})How are VWAP, TWAP and volume distribution calculated?
VWAP uses the typical price of each candle, (high + low + close) / 3, weighted by volume and accumulated over time. TWAP in the skill is a simple running average of closing prices. Volume distribution gives each candle's share of total volume and can be summed into hourly buckets.
These are standard, simple formulas. They describe the data you provide and do not predict price moves.
typical_price = (df["high"] + df["low"] + df["close"]) / 3
df["vwap"] = (typical_price * df["vol"]).cumsum() / df["vol"].cumsum()Can minute-analysis run minute-level backtests?
The SKILL.md describes setting an interval such as 5m in a config.json file and using a backtest tool, with an annualization factor inferred from the data source and interval. However, the bundled example_signal_engine.py states that it supports analysis output only and that the backtest engine supports daily data only. These two statements conflict, and the code has not been independently validated.
Treat backtest support as unconfirmed. Use the skill for data retrieval and indicator analysis, and check any backtest setup against the wider Vibe-Trading system before drawing conclusions.
- Keep the transaction cost lower for minute strategies (the source suggests 0.05% rather than 0.1%) because intraday trading is frequent.
- Watch for empty results from Tushare if your score is below 2000.
What safety and limits should you know about?
This skill runs scripts and uses the network. It fetches data from okx.com and, optionally, Tushare and yfinance. Review the scripts before running them, and only run them in an environment you trust.
The skill analyses historical data. It does not place orders or trade live. Results depend on external services, which apply rate and row limits, so a request can fail or return less data than expected. Timestamps are in milliseconds, so check that dates are converted correctly before reading results.
- No live trading or order execution
- Data depends on internet access and third-party rate and row limits
- Tushare minute data returns empty if the account score is below 2000
What people use it for
- Compare intraday VWAP and TWAP for a crypto pair against the price path to see where trades clustered.
- Break down BTC-USDT trading volume by hour to find the busiest parts of the day.
- Pull US or Hong Kong stock minute candles from yfinance to study the volume pattern around the open.
- Build a data-fetching base for minute-level strategy research, reviewing and validating the code yourself first.
Frequently asked questions
Is the minute-analysis agent skill free?
The skill is published under the MIT licence, as stated on this page. The data sources have their own terms: OKX public candle data needs no API key, yfinance is free and needs no key, and Tushare minute data requires an account with a score of at least 2000.
What do I need to install minute-analysis?
You need an agent that supports skills, Python with pandas, numpy and requests, and internet access to okx.com. Tushare and yfinance are optional, depending on the markets you want.
Where do I put the minute-analysis skill folder?
Unzip the downloaded ZIP into ~/.claude/skills/ for personal use, or into a project's .claude/skills/ folder. In the Claude apps you can instead upload the ZIP under Settings, Capabilities, Skills.
Why does my Tushare request return no data?
According to the skill's notes, Tushare minute endpoints require a score of at least 2000. If the score is lower, the API returns empty data, so check your account score first.
Can I use minute-analysis for live trading?
No. The skill retrieves and analyses historical data and does not execute orders. Its backtest support is also contradictory in the source, so do not base trading decisions on unverified backtests.
How does minute-analysis compare with other data tools?
The source does not compare it with other tools. Its distinguishing features are that it is packaged as an agent skill, covers OKX, Tushare and yfinance in one place, and includes ready-made VWAP, TWAP and volume formulas. Check the code and its outputs before choosing it over alternatives.
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 minute-analysis if you want an agent to fetch intraday candles for crypto or stocks and calculate VWAP, TWAP and volume breakdowns, and you are comfortable reviewing code. Pick something else if you need live trading, validated backtests or a service with guaranteed data access, since the skill's backtest claims conflict and its data depends on external services.
Download the skill