Skip to content

kdhiraj152/trader-skill

v2.1.0MIT

12 skills for day trading: scan thousands of charts, track breaking news, time every entry and exit, replay every outcome before you risk a dollar, and dig into any stock: options, shorts, dark-pool flow, insider trades and the macro numbers.

The Trader skill

tests

12 skills for day trading: scan thousands of charts, track breaking news, time every entry and exit, replay every outcome before you risk a dollar, and dig into any stock: options, shorts, dark-pool flow, insider trades and the macro numbers.

Free, MIT, no signup, no API key. The tools are plain Python that read free public data: Yahoo Finance charts, Nasdaq Trader's symbol lists, SEC EDGAR and the House Clerk. One optional piece goes further: run a local OpenBB server (free, no key) and the Research Desk reads options chains with the greeks, short interest, dark-pool volume, insider trades and macro series through it.

One scans thousands of charts for setups. Another tracks breaking news on every stock you watch. One watches a setup bar by bar and tells you the moment to get in, and another sets your stop and your target before you do. And the craziest one is the Time Machine: before you risk a dollar, it replays the rest of the session 10,000 times from the stock's own recent bars, plus every past day the same setup fired, and tells you the honest odds.

And no trade goes through until you say yes. Nothing in this pack can place an order on its own. Paper trading is the default.

Install

Any agent that loads SKILL.md folders can run this. Four ways in, pick the one that fits how you work:

npx, if you have Node. One command, no paths: it copies the 12 skill folders where agents read them (your project's skills/ folder if there is one, else ~/.agents/skills, created on the spot; --dir to choose yourself):

npx trader-skill

pipx or pip, if you have Python. Same installer, same result:

pipx install trader-skill
trader-skill install

(pip install trader-skill works too; trader-skill list shows what ships, and --skills trader-scanner,trader-news installs a few instead of all 12.)

Clone and copy, the old way. No install step you don't control:

git clone https://github.com/KDhiraj152/trader-skill.git
cp -r trader-skill/trader-* <your agent's skill folder>/

Paste it into the agent. No skill loader, no terminal? Copy the block below and paste it at the top of a chat with your agent; it does the rest. Or open any single SKILL.md in this repo, copy the whole thing, and paste it the same way: the agent picks it up as a mode, and you lose only the Python tools, not the method. The paste for trader-budget or trader-risk alone will save your account.

Install the Trader skill pack for me. Steps:
1. Run: npx trader-skill
   (no Node? use: pipx install trader-skill && trader-skill install)
2. Read every SKILL.md in the folder it names and treat them as your skills.
3. The skills write their state into a trader/ folder in this project.
4. Start with /trader-scanner to build a watchlist, then /trader-news,
   /trader-entry, /trader-exit, /trader-timemachine, /trader-budget,
   /trader-risk before any order. Paper trading only by default; no order
   is ever placed without my explicit yes.

As a plugin. The repo root has plugin.json and marketplace.json in the neutral agent-plugins.org shape. Agents that read that manifest can install the whole pack straight from this repo.

The tools need Python 3.8 or newer and nothing else: plain standard-library Python, no pip dependencies. The SEC tools (news filings, the analyst) ask you to identify yourself: put your name and email in trader/config.json ("sec_user_agent": "Your Name you@example.com") or in the SEC_USER_AGENT environment variable.

The Research Desk needs a running OpenBB server: install OpenBB, run openbb-api from its environment, and the desk works. Without it, that skill says so and the other eleven run as before.

The twelve

commandjobwhat it does
/trader-scannerChart ScannerScans thousands of charts for setups: opening-range breakouts, flags, oversold bounces.
/trader-newsNews MonitorTracks breaking news on every stock you watch and flags what moves it.
/trader-entryEntry TimerWatches a setup bar by bar and tells you the moment to get in.
/trader-exitExit PlannerSets your stop and your target before you enter, so you know when to get out.
/trader-timemachineTime MachineReplays every possible outcome of a trade before you risk a dollar: 10,000 simulated sessions plus every past time the setup fired.
/trader-riskRisk ManagerChecks every trade against your limits and stops anything over them.
/trader-budgetBudget ManagerSizes every position to your budget and your max loss per trade.
/trader-brokerOrder ExecutorSends approved orders to your broker, paper trading by default, never without your yes.
/trader-alertsPrice AlertsRings when a price, a stop or the opening bell hits.
/trader-analystEarnings AnalystReads earnings reports and sums up what changed.
/trader-pelosiPelosi TrackerReads Congress's stock-trade disclosures the minute they are filed.
/trader-researchResearch DeskOptions chains with the greeks, short interest, dark-pool volume, insider trades, filings, the earnings calendar and macro series, from a local OpenBB server.

How to use it

Everything lands in a trader/ folder in your project, and each skill reads what the last one wrote:

scanner -> news -> entry -> exit -> timemachine -> budget -> risk -> broker   (one trade)
alerts: stop, target, flat by the close                                       (while it is open)
news / analyst / pelosi                                                        (on a schedule)
research: shorts, options, insiders, macro                                      (before and after)

An example, the trade in the reel: Tesla, Monday October 19 2026, real data.

  1. /trader-scanner at 10:15 AM ET: of the 231 charts in the reel's scan, 42 had closed above their 15-minute opening range. TSLA is one: range 354.47 to 362.18, the 10:10 candle closed at 364.18, 0.55% over the top.
  2. /trader-news: the morning's TSLA headlines, in New York time ("Tesla-SpaceX Merger Mania Strikes Wall Street Ahead of Earnings", 8:04, tagged earnings and deal). Tesla reports this week: a day trade is out before the close, so it is not exposed to the report.
  3. /trader-entry watched TSLA bar by bar. At 10:12 it said waiting (the 10:10 bar was still forming). At 10:15 it said triggered: the first 5-minute close above the range, 364.18.
  4. /trader-exit set the plan before the entry: stop at the range's midpoint, 358.32, so 5.86 a share at risk; target one times that risk, 370.04; flat by 3:55 PM either way.
  5. /trader-timemachine replayed the rest of the day 10,000 times, each 5-minute candle drawn from TSLA's own candles at that time of day over the previous 59 sessions (seed 19, so anyone can repeat it): target first 29.4%, stop first 25.5%, still open at the close 45.1%. It also re-ran the same rule on those 59 real sessions: it fired 33 times, 13 hit the target, 15 the stop, 5 neither. Its verdict: close to a coin flip. Which is exactly why you replay it first.
  6. /trader-budget with a $5,000 account and a $50 max loss per trade: 50 / 5.86 = 8 shares ($2,913.44). If the stop fills at 358.32, the loss is $46.88. It writes one bracket order: buy 8 TSLA, stop-loss 358.32, take-profit 370.04.
  7. /trader-risk checks it against your limits: the max loss, the daily loss limit, trades per day, reward to risk, the entry window, position size and cash. PASS or BLOCK, with the reason.
  8. /trader-broker shows you the ticket in plain words. You type yes for that exact order. Your agent calls your brokerage connector (Alpaca's MCP server, paper account by default), and its "Allow once / Deny" approval asks you one more time.
  9. /trader-alerts rings on the stop, the target and at 3:55 PM to flatten whatever is still open.

The tools

All standard-library Python. The data tools read public endpoints and are polite about it.

python3 trader-scanner/scan.py universe                                # ~5,900 US stocks, Nasdaq Trader
python3 trader-scanner/scan.py daily                                   # pre-market setups, whole universe
python3 trader-scanner/scan.py orb --at 10:15                          # opening-range breakouts so far
python3 trader-news/news.py --tickers TSLA --hours 12                  # headlines, 8-Ks, today's move
python3 trader-entry/entry.py check TSLA --setup orb                   # waiting / triggered / expired
python3 trader-entry/entry.py watch TSLA --every 30                    # bar by bar until it triggers
python3 trader-exit/exit.py plan --stop or-mid --target 1R --write     # -> trader/plan.json
python3 trader-timemachine/timemachine.py run --write                  # 10,000 replays + the history
python3 trader-budget/size.py --write                                  # -> trader/orders.json
python3 trader-risk/check.py                                           # PASS / BLOCK per order
python3 trader-broker/ticket.py show                                   # the tickets, READY or not
python3 trader-alerts/alerts.py add --from-plan                        # stop, target, flat by close
python3 trader-alerts/alerts.py check                                  # rings what hit, once
python3 trader-analyst/facts.py TSLA                                   # quarters, growth, EPS
python3 trader-research/research.py shorts TSLA                        # short interest, days to cover
python3 trader-research/research.py chains TSLA --expiration 2026-11-20
python3 trader-research/research.py insider TSLA                       # Form 4 trades
python3 trader-research/research.py calendar --days 7                  # who reports this week
python3 trader-research/research.py macro FEDFUNDS                     # a macro series
python3 trader-pelosi/ptr.py index --member pelosi                     # her PTRs this year
python3 -m unittest discover -s tests -v

The tests run offline, on fixtures cut from real data: TSLA's 5-minute bars for the 60 days to October 19 2026, the 10:15 AM opening-range scan, Yahoo's TSLA headlines that morning, the House Clerk's 2026 index, real PTR text and real XBRL facts. They reproduce every number in the example above.

Fine print

Not financial advice. This is software that reads public data and does arithmetic. It does not know your situation and it cannot see the future. Most people who day trade lose money. You can lose money. Decide for yourself, and talk to a licensed professional if you need advice.

No trade goes through until you say yes. Only /trader-broker can lead to an order, and only after /trader-risk passes that exact order and you type yes for it in the chat. Set your brokerage tool to "Needs approval" in your agent's connector settings, never "Always allow", so its "Allow once / Deny" prompt is a second check. No options, margin, short selling or leverage unless you turn them on.

Paper first. "paper": true is the default in config.example.json, and Alpaca's MCP server defaults to its paper account. Keys live in environment variables, never in the chat.

The Time Machine is a simulation, not a prediction. It is built from the stock's own last 60 days of 5-minute bars. The future can do things those 60 days never did: an earnings report, a halt, a headline. Treat its odds as a reality check, never as a promise.

A stop is a plan, not a guarantee. In a fast market or across a gap, a stop order can fill well past its price. The max loss shown is the loss at the stop price.

The data is free and unofficial. Yahoo Finance's chart endpoint is not an official API: it can lag, change or rate-limit you. The scanner and the entry timer read 1-minute or 5-minute bars, not every tick. Check anything that matters against your broker before you act on it.

The research data is delayed, deliberately. The Research Desk reads a local OpenBB server, and its sources are not live feeds: CBOE quotes and option chains are 15 minutes or more behind, FINRA short interest comes twice a month, dark-pool volume weekly, and fails to deliver about a month behind. None of it is a signal by itself; it is positioning context to read next to the filings and the tape.

Day-trading rules. FINRA removed the $25,000 pattern-day-trader minimum on June 4, 2026 (Regulatory Notice 26-10), and brokers may phase the change in until October 20, 2027, so yours may still apply the old rule. If it does, set "pdt_guard": true and /trader-risk counts your day trades.

Congress's trades are public, not inside information. Members must disclose trades over $1,000 within 45 days (the STOCK Act). The Pelosi skill reads those public filings the minute they post.

Taxes. Day-trading gains are short-term gains, and frequent trading makes wash sales likely. Nothing here is tax advice.

When something goes wrong

Most problems here are one of five things.

A data source stopped answering. Yahoo's chart endpoint is not an official API: it can rate-limit you (empty or missing bars), change its JSON shape, or just lag. If a scan comes back with no usable charts, wait a few minutes and try again before assuming every stock died. The Nasdaq Trader lists, SEC EDGAR and the House Clerk are more reliable, but they all publish on their own schedule: the universe is rebuilt when you run universe, PTRs appear when the Clerk posts them.

SEC tools return nothing and say why. EDGAR asks every automated user to identify itself. Put your real name and email in trader/config.json ("sec_user_agent": "Your Name you@example.com") or set the SEC_USER_AGENT environment variable. Without it the news tool still shows headlines and just skips the filings.

A guard refused your command and it looks like an error. It is the tool checking the data. The common ones: a price quote older than 15 minutes is not used ("Sized at the plan's entry"); a plan written on an earlier day is refused (make a new one with /trader-exit); a risk check from an earlier day makes the ticket "NOT READY" (run check.py again); an order without a stop is blocked while "require_stop" is on, which is the default. Read the reason: it names the file that is stale or the limit that was hit.

The broker never receives anything. That is the design. /trader-broker only prints the ticket and the exact tool-call JSON; your agent's brokerage connector is what actually places it, after you type yes for that one order. If nothing happens after you say yes, check that the connector is set up (mcp_servers.example.json shows the shape), that its tool permission is "Needs approval" rather than "Always allow", and that "paper": true is pointing at the paper account you expect.

A PTR PDF will not parse. /trader-pelosi falls back to printing the PDF path so you can open it yourself. For automatic text extraction it needs poppler: brew install poppler (macOS) or your package manager's equivalent, then ptr.py fetch again.

The research desk refuses to run. It talks to a local OpenBB server and there is none up: start it with openbb-api --port 6900 from your OpenBB environment, or point the desk elsewhere with openbb_base_url in trader/config.json. The error names both. Agents that speak MCP can connect openbb-mcp instead; the shape is in trader-research/mcp_servers.example.json.

Files

plugin.json          the neutral agent-plugins.org manifest (whole pack installs as one plugin)
marketplace.json     same, for agents that read a marketplace list
pyproject.toml       the pip package: pipx install trader-skill, then `trader-skill install`
package.json         the npm package: npx trader-skill
bin/cli.js           the npm installer (plain Node, no dependencies)
installer/           the pip installer: copies the bundled skill folders out
trader-scanner/     SKILL.md, scan.py
trader-news/        SKILL.md, news.py
trader-entry/       SKILL.md, entry.py
trader-exit/        SKILL.md, exit.py
trader-timemachine/ SKILL.md, timemachine.py
trader-risk/        SKILL.md, check.py
trader-budget/      SKILL.md, size.py, config.example.json, targets.example.json
trader-broker/      SKILL.md, ticket.py, mcp_servers.example.json
trader-alerts/      SKILL.md, alerts.py, alerts.example.json
trader-analyst/     SKILL.md, facts.py, metrics.example.json
trader-pelosi/      SKILL.md, ptr.py
trader-research/    SKILL.md, research.py, mcp_servers.example.json
portfolio/          the v1 basket: copycat, screener, tax, rebalance, plus the same pack as above
tests/              unittest suite + offline fixtures, for both the pack and the basket

MIT. Made by K Dhiraj.