$29 · Instant download
A trading agent on your own Alpaca account, with a written mandate, real guardrails and a kill switch — placing paper orders by the weekend. Not a course about trading.
[3/4] Placing a market buy: 1 x SPY [4/4] Confirming order id 8f2c1a04-… status filled filled 1 @ 601.22 Done. Your agent placed a trade and you have a fill to prove it. Check it yourself: app.alpaca.markets/paper/dashboard/orders
That is module 1, start to finish, in about twenty minutes. Paper account — no real money anywhere in it.
One payment. Instant download. Six modules.
If you came from the Alpaca session, this is the first thing you want to know.
That design has strategy agents proposing trades, a critic checking a proposal is structurally valid without judging whether it will work, a deterministic risk guard with no model in it, and a human deciding before anything executes.
This kit builds the desk, the critic, the risk guard and the gate. It runs as many agents as you register, drops their contradictions instead of netting them, refuses to stack their agreements, and re-checks your limits after every approval — because five orders that each pass individually will still break a daily cap together.
The agents that form a view are yours to write. The kit ships one, and it abstains. That is the line between a toolkit and advice, and it is deliberate: a product that hands you a momentum agent is telling you what to buy. What is hard here was never the opinion — it is everything that has to be true before an opinion is allowed to spend money.
Six modules. Every one ships something that runs, not something you read.
Account, keys, tooling, and one confirmed order end to end.
Several agents proposing at once, and limits none of them can argue past — as configuration, not a prompt they can be talked out of.
Every Alpaca account opens able to spend more than you put in, and nobody asks you first. Plus a daily loss limit and a stop that survives a reboot.
The part every retail project skips and every professional desk runs. You will know it when you get there.
Six drills against your own system before it sees real money. Each has a right answer, written down in advance.
It reads what your system has actually done and answers from evidence. It is written to tell most people to wait.
Two of these are why it's legal. All three are why it's worth $29.
No strategy, no signals, no ticker. FIRST_TRADE_SYMBOL ships blank and stays blank. A kit that hands you a symbol is teaching you the symbol is the interesting part. It isn't.
Your keys, your machine, your account, your decisions. Nothing is executed on your behalf and no discretion is ever held.
Not for legal reasons — for credibility ones. Every competitor leads with a suspiciously good one. If you've been burned by that, you already know why there isn't one here.
Three ideas, applied everywhere.
A setting that doesn't exist can't be got wrong at 1am. Module 1 has no path to real money, because the setting was never written. The broker cannot be constructed without a mandate. The kill switch is a file, so it survives the restart that would otherwise undo it.
Fail closed, always. A price it cannot read is unmeasurable, not free. A corrupt stop-file still stops. A check that could not run is never recorded as a check that passed.
Never fix silently. When something is wrong it halts and tells you, rather than quietly correcting itself and destroying the evidence that anything happened.
You can read Python and you've wired up an API before. You've watched someone build an AI trading agent on a stage and thought I could do that, properly. You'd rather own the machine than rent a signal.
You do not need to know anything about trading strategy, because this kit doesn't have any.
Requires a free Alpaca account. Python 3.10+.