Why it mattersA skill pack is becoming a real way to release software. One repo gives a retail trader a working system that reads free public data.
Most people who want to try day trading do not have a safe way to practice. Day Trader Skill is a free pack of 11 Claude skills by Jake Schincariol. The pack scans thousands of charts and times entries bar by bar. Before any order is sent, it replays the rest of the session 10,000 times. The GitHub repository was created on 20 October. Trendshift placed it 23rd on its daily list on 25 October. The repo has 1,526 stars so far.
Schincariol says all 11 tools are plain Python. They read free public data. Yahoo Finance supplies the charts. Nasdaq Trader supplies the symbol list. SEC EDGAR supplies the filings. The House Clerk supplies the Congressional trade disclosures. There is nothing to install with pip. The code is MIT licensed.
What the eleven skills do
Each skill does one job. It writes its output to a trader/ folder. The next skill reads what the last one wrote. The chain is scanner, news, entry, exit, Time Machine, budget, risk, broker. Three other skills run on a schedule. One is a Pelosi tracker that reads Congress's public stock trade disclosures the minute they are filed.
The scanner looks across about 5,900 US stocks. It finds opening-range breakouts, flags and oversold bounces. The entry timer watches a setup bar by bar. It says waiting, triggered or expired. The exit planner sets a stop and a target before the trade opens. The budget sizer picks a share count that keeps the loss under a chosen amount.
How the Time Machine makes its number
The Time Machine runs 10,000 simulated sessions. Each simulated candle is drawn from the stock's own 5-minute bars. The source is the last 59 sessions at that time of day. The skill also re-runs the same rule on those 59 real sessions. It reports how many times the target hit first, the stop hit first, or neither.
Schincariol gives a worked example in the README. The example is a Tesla trade on 20 October. The Time Machine reported target first 29.4 percent, stop first 25.5 percent, still open at the close 45.1 percent. The 59-day backtest fired the rule 33 times. 13 hit the target, 15 the stop, 5 neither. His verdict in the README is "close to a coin flip. That is the point of looking before you risk a dollar."
The limits the author states
The README calls out several limits. The Time Machine is a simulation. The future can do things the last 60 days never did. A stop is a plan for an exit price. In a fast market a stop order can fill well past that price. Yahoo Finance's chart endpoint is a public web endpoint without a support contract. It can lag or rate-limit a caller.
The order-execution skill never places a trade on its own. The reader has to type yes for each ticket in the chat. Claude's own tool-permission card then asks a second time. Paper trading is the default. The broker connector supports Alpaca's MCP server.
Day Trader Skill is a sequel to Schincariol's earlier Trader skill. He has released 11-skill packs for LinkedIn posts, YouTube videos, cloning any app, and validating a business. One repository with a SKILL.md file in each folder is becoming a product shape of its own. For a buyer of software, that changes who the vendor is. The vendor becomes the author of a 500-line Python tool.
Source
This item was written by an AI system from the linked source. Reveneau is responsible for what it publishes.
Get AI News in your inbox
New developer tools, model and agent releases, and how teams are actually using them to release software. Short, and only when there is something worth reading.