Build a Safer Day-Trading Workflow with Claude’s Day Trader Skills

What the Day Trader project does

The Day Trader skill is an MIT-licensed collection of 11 Claude skills for researching and managing day trades. It can scan thousands of US stock charts, monitor public news and filings, identify entry signals, plan exits, simulate possible outcomes, size positions, enforce risk limits, prepare broker orders, and monitor alerts.

The included Python tools use free public data from sources such as Yahoo Finance, Nasdaq Trader, SEC EDGAR, and the House Clerk. They require no package installation, signup, or API key. Python 3.8 or newer is required.

Important: The project cannot independently place a trade. Only the broker skill can lead to an order, and it requires the risk check to pass, your explicit confirmation, and the brokerage connector’s approval prompt. Paper trading is the default.

Key features and commands

• /trader-scanner: Scans charts for opening-range breakouts, flags, and oversold bounces.
• /trader-news: Monitors headlines, SEC filings, and price-moving developments for watched stocks.
• /trader-entry: Checks a setup bar by bar and reports whether it is waiting, triggered, or expired.
• /trader-exit: Defines a stop, profit target, and time-based exit before entry.
• /trader-timemachine: Runs 10,000 simulated sessions and examines historical occurrences of the same setup.
• /trader-budget: Calculates position size from account budget and maximum acceptable loss.
• /trader-risk: Returns PASS or BLOCK after checking the proposed order against configured limits.
• /trader-broker: Displays an order ticket and can send an explicitly approved order through a brokerage connector.
• /trader-alerts: Watches stop, target, price, opening-bell, and end-of-day conditions.
• /trader-analyst: Reads earnings information and summarizes changes in company performance.
• /trader-pelosi: Reads public congressional stock-trade disclosures when they are posted.

Installation options

Install directly through Claude

Paste the repository URL and installation instruction into Claude:

https://github.com/Jakeschincariol/day-trader-skill

install skill

Install as a Claude Code plugin

Add the repository as a plugin marketplace and install the plugin:

/plugin marketplace add Jakeschincariol/day-trader-skill
/plugin install day-trader-skill@day-trader-skill

Claude Code namespaces plugin commands. For example, the scanner appears as /day-trader-skill:trader-scanner rather than /trader-scanner.

Install plain global commands

Clone the repository and copy its skill directories into your global Claude skills directory:

git clone https://github.com/Jakeschincariol/day-trader-skill.git
cp -r day-trader-skill/trader-* ~/.claude/skills/

To keep the skills local to one project, copy the same directories into that project’s .claude/skills/ directory instead.

Use a skill without Claude Code

You can paste the contents of an individual SKILL.md file at the beginning of a chat. Claude can then follow that skill’s method as a conversational mode, although the associated Python tools will not be available.

Configure the environment

Confirm that Python 3.8 or newer is installed:

python3 --version

The tools use only Python’s standard library, so there is nothing to install with pip.

The SEC-dependent news and analyst tools require an identifying user agent. Add your name and email to trader/config.json:

{
  "sec_user_agent": "Your Name you@example.com"
}

Alternatively, define the SEC_USER_AGENT environment variable:

export SEC_USER_AGENT="Your Name you@example.com"

Keep brokerage credentials in environment variables rather than pasting them into a chat. The example configuration uses "paper": true, and Alpaca’s MCP server defaults to its paper account.

Understand the workflow

Each skill writes its output into a trader/ directory, allowing the next skill to consume the previous result. A typical single-trade workflow is:

scanner -> news -> entry -> exit -> timemachine -> budget -> risk -> broker

While a position is open, the alerts skill can watch its stop, target, and required end-of-day exit. The news, analyst, and congressional-disclosure skills can also run on a schedule.

Step-by-step basic usage

1. Build or scan the stock universe

Download the Nasdaq Trader symbol universe, which contains roughly 5,900 US stocks:

python3 trader-scanner/scan.py universe

Run a whole-universe scan for pre-market setups:

python3 trader-scanner/scan.py daily

To look for opening-range breakouts observed by 10:15, run:

python3 trader-scanner/scan.py orb --at 10:15

2. Check news and filings

After selecting a candidate, inspect recent headlines, 8-K filings, and the day’s move. This example requests the previous 12 hours of TSLA information:

python3 trader-news/news.py --tickers TSLA --hours 12

News context matters because a technically valid setup can behave differently around earnings, mergers, regulatory developments, or other market-moving events.

3. Validate the entry

Check whether TSLA’s opening-range breakout setup is waiting, triggered, or expired:

python3 trader-entry/entry.py check TSLA --setup orb

To continue monitoring bar by bar every 30 seconds until the setup triggers, use:

python3 trader-entry/entry.py watch TSLA --every 30

The entry tool reads one-minute or five-minute bars rather than every market tick. Confirm critical prices with your broker before acting.

4. Define the exit before entry

Create a plan using the opening range’s midpoint as the stop and one unit of risk as the target:

python3 trader-exit/exit.py plan --stop or-mid --target 1R --write

The --write option saves the result to trader/plan.json. Planning the stop, target, and end-of-day exit in advance prevents later tools from relying on an undefined exit strategy.

5. Simulate and back-check the setup

Run the Time Machine after creating the plan:

python3 trader-timemachine/timemachine.py run --write

The tool generates 10,000 simulated sessions from the stock’s own recent five-minute bars and also checks historical sessions in which the same rule fired. It reports outcomes such as target first, stop first, or still open at the close.

The Time Machine is a simulation, not a prediction. Its source window cannot represent every future halt, gap, earnings surprise, or headline. Use its output as a reality check rather than a promise.

6. Calculate position size

Once entry and stop prices are known, generate a position size and proposed order:

python3 trader-budget/size.py --write

The budget tool uses the configured account size and maximum loss per trade. For example, the README’s TSLA walkthrough uses a $5,000 account, a $50 loss limit, and $5.86 of planned risk per share. Dividing 50 by 5.86 produces a position of eight shares, with a planned stop-price loss of $46.88.

The generated bracket order is written to trader/orders.json.

7. Run the risk gate

Check every proposed order before presenting it to the broker:

python3 trader-risk/check.py

The risk manager evaluates limits including maximum trade loss, daily loss, trades per day, reward-to-risk ratio, entry window, position size, and available cash. It returns PASS or BLOCK with a reason.

8. Review the broker ticket

Display the prepared ticket:

python3 trader-broker/ticket.py show

The broker step shows the order in plain language and indicates whether it is ready. It does not silently submit the order. You must approve that exact order in the chat, and the brokerage connector should present a second Allow once or Deny confirmation.

9. Add and check alerts

Create alerts from the saved plan:

python3 trader-alerts/alerts.py add --from-plan

Check whether the stop, target, or time-based exit has been reached:

python3 trader-alerts/alerts.py check

The alert checker rings each matched condition once. A typical plan includes flattening any remaining position near the close, such as 3:55 PM.

Additional research tools

Review earnings facts

Request quarterly facts, growth, and earnings-per-share information for TSLA:

python3 trader-analyst/facts.py TSLA

Index public congressional disclosures

Index Pelosi-related periodic transaction reports for the current year:

python3 trader-pelosi/ptr.py index --member pelosi

These reports are public disclosures, not inside information. Under the STOCK Act, covered transactions over $1,000 may be reported as late as 45 days after they occur.

Test the installation offline

Run the included unit tests from the repository root:

python3 -m unittest discover -s tests -v

The tests use offline fixtures based on real data, including TSLA five-minute bars, a 10:15 opening-range scan, Yahoo Finance headlines, House Clerk records, PTR text, and XBRL facts. They reproduce the numerical results documented in the project’s example.

Advanced tips for safer use

1. Start with paper trading. Leave "paper": true enabled until you have validated the complete workflow and understand every generated file.
2. Preserve both approval gates. Set the brokerage connector’s tool permission to Needs approval, never Always allow.
3. Inspect intermediate files. Review trader/plan.json and trader/orders.json before running later stages. The chained workflow is convenient, but each stage should still be verified.
4. Use reproducible simulations. Record the Time Machine’s seed when comparing strategy changes so the same simulated sequence can be repeated.
5. Treat stop losses as estimates. A stop order can fill beyond its specified price during a gap or fast market. The displayed maximum loss assumes execution at the stop price.
6. Verify unofficial data. Yahoo Finance’s chart endpoint may lag, change, or rate-limit requests. Compare important prices and account details with your broker.
7. Configure applicable day-trading controls. Brokers may apply different day-trading requirements during regulatory transitions. If your broker still uses the older pattern-day-trader rule, enable "pdt_guard": true so the risk tool counts day trades.
8. Avoid unplanned leverage. Options, margin, short selling, and leverage remain disabled unless you explicitly enable them.
9. Consider taxes. Day-trading profits are generally short-term gains, and frequent transactions can create wash sales. The project does not provide tax advice.

Limitations and risk warning

This software reads public data and performs calculations; it does not know your financial circumstances or predict future prices. Most people who day trade lose money. Simulated probabilities, historical tests, entry signals, and planned stops do not guarantee a profitable trade or a specific execution price.

Use the project for research and paper trading, independently verify important information, and consult a licensed professional when you need personalized financial or tax advice.

Conclusion

The Day Trader skill turns a trade idea into a structured, reviewable sequence: scan, investigate, validate, plan, simulate, size, check risk, request approval, and monitor. Its strongest safeguards are the explicit risk gate, paper-trading default, and mandatory human confirmation. Explore the source and updates in the Day Trader skill GitHub repository.