Day Trader Skill Review: What the 11-Skill Toolkit Actually Does
Day Trader Skill is an open-source package of 11 Claude skills and Python helpers. Its documented workflow scans charts, inspects news, plans an entry and exit, sizes a proposed trade, checks risk rules, and displays a broker ticket. One outside article tested selected offline behavior. This review shows what those results support, where the source reports problems, and what to inspect before trying the tools yourself.
What is in the package?
The source repository groups the skills into distinct folders with SKILL.md instructions and Python helpers. Scanner, News, Entry, Exit, Time Machine, Budget, Risk, and Broker form its example trade path. Alerts monitors prices and times; Analyst reads SEC company facts; Pelosi reads public House disclosure reports. The repository's plugin manifest identifies its own version as 2.0.0. That is the source software's version, not this guide site's V1.0.
The source describes Python 3.8 or newer and no additional pip dependency for its basic tools. Data and brokerage behavior still depend on external services, local configuration, and Claude permissions. Reading a Skill instruction file in a normal chat does not by itself run its Python helper. The site's installation section distinguishes a Claude Code install from chat-only reading and local folder copy.
What has been checked independently?
Nanoflow's November 2026 article reports a limited offline test: the project's 315 tests passed in its environment, its saved chart scan matched a separate count of 42 out of 231 usable charts, and a comparison across 428 historical setups on 20 large stocks found the simulation's target-first share was 38.9% versus 33.9% in that observed sample. The article says it used no broker, account, or real money. These are the author's results, not a hands-on test by this site. An offline test of selected inputs checks consistency and arithmetic; it does not measure current market execution, future strategy performance, or profitability.
| Evidence | Supports | Does not support |
|---|---|---|
| Source repository and tests | Intended commands, data flow, example behavior | Live reliability in every user environment |
| Nanoflow offline test | Selected offline behavior and historical comparison in its stated sample | Future results or profitable trading |
| GitHub issue reports | Specific user-described failure cases | A blanket finding that every run fails |
How should the Time Machine numbers be read?
The Time Machine source instructions say the tool creates 10,000 simulated session paths from recent five-minute bars and separately checks past occurrences of the setup. In the project's dated October 19, 2026 TSLA example, the simulated path shares are 29.4% target first, 25.5% stop first, and 45.1% neither by the close. The historical replay has 33 rule occurrences: 13 target first, 15 stop first, and 5 neither. These are different calculations with different denominators. The recent bars cannot contain a future halt, unexpected news, or a new market regime.
A GitHub user report, Issue #19, describes a case where the tool still displayed simulated odds after later historical bars had already resolved the outcome. The issue was open when checked on November 5, 2026. The report is a reason to inspect the exact data window and code behavior, not a conclusion that every simulation is wrong.
What about risk checks and broker execution?
The Risk skill checks a proposed order against configured constraints such as planned loss, daily limits, entry time, and position size. A PASS is a rule result over the available inputs. Issue #32 reports a market-order case in which a fresh price had crossed the planned stop while the risk check still passed. The issue included a reproduction and proposed patch and was open on November 5, 2026. This is a user report; the cited material does not establish that a fix was released. Before using a broker ticket, inspect its current quote, planned stop and exact order details.
The Broker skill describes paper mode as default and requires the exact order to pass risk rules, a typed user approval, and a Claude tool-permission confirmation when set to Needs approval. These conditions are in source instructions; they are not a guarantee that a user will configure a connector correctly. This guide has no broker connection and cannot submit a trade. A planned stop may fill at a worse price in a fast market or gap.
Who could use it?
The pack is best suited to readers who understand basic day-trading risk and want to inspect the scripts behind each result. Start with a historical or paper example, then check the data timestamp and rule inputs at every step. Beginners can use the guides to learn the workflow; they still need to understand orders, risk limits and tool permissions before using a broker connection. The repository and limited outside test provide no evidence of an automatic profitable strategy.
Verdict
Day Trader Skill has an inspectable source repository and a documented paper-trading workflow. The available outside test covers selected saved-data behavior; it does not establish live execution reliability or a profitable strategy. Read the source, follow the local paper-trading walkthrough if you have cloned the repository, and check the risk guide before using a broker connection.
Common questions
Did this site run its own trading test? No. We inspected the public source and linked independent and community reports.
Is 38.9% a success rate? It is the simulated target-first share reported for one historical comparison method, not a future win rate or return.
Does an open GitHub issue prove the tool is unsafe in every case? No. It documents a specific reported case that deserves a current-code check before relying on the affected path.
Sources
- Original README, plugin manifest, and linked Skill files; checked 2026-11-05 against
maincommit3d0f4ef. - Nanoflow independent article; third-party methodology and figures, checked 2026-11-05.
- GitHub Issue #19 and Issue #32; user reports, both open at the check.