Limitless Trader Lab · graduation day
Day 7 of 7 · Cohort intensive · 60–90 minutes
Go live + graduation.
Six days of building. Today you flip the switch, one real live trade and one deliberate kill-switch test, then you join the alumni and the self-guided phase begins.
Today you graduate
You’ll learn how to flip DRY_RUN=false, let your deployed agent place one real $1–$5 live trade on Limitless, then deliberately trip and reset the kill-switch. This will help you walk out of the cohort with proof the wire works both ways, one live trade and one clean halt, which is the exact graduation bar that unlocks the Trader Lab Alumni role and the self-guided road ahead.
Section 01
Why go live small.
Today’s graduation bar isn’t “made money”, it’s “made one trade and stopped one trade.” The live trade proves the agent can transact for real. The kill-switch test proves you can stop it for real. Both matter equally. The trade is small ($5 max) on purpose: today’s lesson is that the wire works end-to-end, not that the strategy is good. Strategy improvement is what the alumni community and the Agents Academy are for.
Place one real trade
One order ID on Limitless, placed by your deployed agent, not a button click. Size is small; the proof is that the wire fires.
Stop one real trade
An agent you can’t stop is a liability. Today you trip the kill-switch on the deployed host, watch it halt, and bring it back.
Cap the blast radius
$5 max position, drawdown halt at -$5, allow-list active, limit orders only. The strategy can’t override the rails.
Section 02
Pre-flight check.
Before you flip DRY_RUN off, walk through every safety rail one more time. Today is the day a missing rail costs real money. If any item below isn’t green, fix it before you flip the switch.
Pre-live checklist
Position cap = $5. Open the Trading skill config and confirm max_order_size_usd: 5. The skill enforces this; the strategy can’t override.
Allow-list active. Trading skill should only place orders on markets in your watchlist. Confirm.
Limit orders only. Market orders are off in the Trading skill config. Confirm.
Drawdown threshold = -$5. Monitor halts the agent if cumulative loss hits -$5. Confirm in monitor/config.yaml.
Kill file path verified. The agent halts when /app/data/kill.flag exists on the deployed volume. Two ways to create it: tap the kill button in your phone panel, or run railway run "touch /app/data/kill.flag" from your laptop. Test both work before flipping the switch.
Section 03
Flip the switch + place one trade.
Set DRY_RUN=false in the host’s env vars. Restart the agent. Watch one full iteration. If the strategy proposes an action this iteration, let it fire. If not, manually trigger one trade via Claude. Either way, the deliverable is one real order on Limitless, placed by the agent, with the agent’s logs.
You are now: flipping DRY_RUN to false
From your laptop, update the env var on Railway and restart the service.
Verify in the next iteration’s log that the agent is in live mode. Look for: {"mode": "live"} in the JSON line.
You are now: waiting for the agent’s first live action
Tail the host logs. The agent runs its scan / evaluate / propose loop. If a signal fires, it places an order. If 10 minutes pass with no signal, manually trigger via Claude:
You are now: confirming the live trade in Limitless UI
Open Limitless UI in another tab. Look at recent orders. You should see one order placed and canceled by your agent. Screenshot it.
If you see multiple orders, your agent is over-firing, trip the kill-switch immediately (next section), then debug.
Section 04
Trip + reset the kill-switch.
The most important deliverable of this entire cohort. An agent you can’t stop is a liability. Today you prove you can stop yours, deliberately, on the deployed host, with the kill-switch you configured on Day 5.
You are now: tripping the kill-switch
Two equally good ways: tap the kill button in your phone panel (writes /app/data/kill.flag via the panel’s API), or create the file from your laptop using railway run. Either way, the agent halts on the next iteration. The path is the same one your Day 5 monitor/config.yaml resolves to via KILL_SWITCH.
Or, for the gesture you’ll actually use day-to-day: pull out your phone, open the panel, tap the kill button, confirm the dialog. The panel POSTs to its own kill endpoint which writes the same /app/data/kill.flag file. Same effect, same halt, no terminal needed.
You are now: confirming the halt
Within one tick (60 seconds), the Monitor skill detects the kill file and halts the loop. Look for the halt message in the logs:
You are now: resetting
Delete the kill file on the deployed volume, then redeploy so the agent starts a fresh loop. Easiest path: tap the kill button in the panel again (it deletes the file via the same endpoint). CLI works too.
If the agent doesn’t resume after the redeploy, check railway logs for the startup line, the fix is usually one of: stale DRY_RUN still pointing at true, missing KILL_SWITCH env var, or a typo in the Procfile from Day 2. Note it for the graduation call, we’ll workshop it.
Section 05
Graduation deliverable.
Two artifacts, one post. The live trade screenshot proves the wire works. The kill-switch log proves you can stop it. Both are required for alumni status. Coach swaps your Trader Lab N cohort role for Trader Lab Alumni within an hour of your post.
Day 7 · Deliverable
Post in #trader-lab.
Screenshot of Limitless UI showing the live trade your agent placed.
Screenshot or log excerpt showing the kill-switch tripped and reset cleanly. Lead the caption with [D7] so the coach catches it on the first scan.
Caption: “[D7] Day 7 done. Live trade placed: [order ID]. Kill-switch tested. I’m an API trader now.”
Coach swaps your role to Trader Lab Alumni within 1 hour of your post. See you on the graduation Google Meet at 7pm local.
Section 06
Graduation call · 7pm local.
One last live call together. Sixty minutes, every cohort member on camera, three rounds. Show your trade. Share your snag. Pick your next strategy. The call is also where the coach hands out the alumni role to anyone who hasn’t posted yet.
Show your live trade
60-second demo: the order ID, the place + cancel log lines, and the strategy that fired (or the manual trigger you used). Coach will queue everyone in cohort order; have your screen ready to share.
Share one snag
The thing that almost stopped you this week, a deploy bug, a signing error, a bad config. Brief: 30 seconds, what broke and what fixed it. The next cohort will benefit from your snag making it into the FAQ.
Pick your next strategy
Coach will walk through the 6 strategies on strategies.html and you’ll claim one for your first 30 days as an alum. Picking out loud creates accountability, alumni use a strategy emoji on their #trader-lab intro post so the cohort can find peers running the same rule.
Section 07
Welcome to the alumni.
Once your deliverable post lands and the coach swaps your role, you stay right where you are: #trader-lab remains your home, but now alongside graduates from every future cohort. The cohort is over; the work is just starting.
You are now: getting the alumni role
Once your [D7] post is up in #trader-lab, coach swaps your @Trader Lab N cohort role for @Trader Lab Alumni within an hour.
The role is permanent, it carries forward through every future cohort. Same channel, new context: you’re on the answering side now.
You are now: introducing yourself as an alum
Drop one short post in #trader-lab: who you are, what your bot does, the strategy you picked on the graduation call. This is how the next cohort finds mentors when they show up.
Suggested template: “[ALUM] Hi, I’m [name], Trader Lab alum (Cohort [N]). My bot trades [strategy] on [market category]. Happy to help with [what you got good at].”
You are now: setting your next milestone
The graduation bar was “made one trade, stopped one trade.” Your next bar is up to you. Most alumni pick one of three:
A 7-day unattended run (the original Day 7 stretch goal); the Agents Academy capstone (production-grade agent); or a strategy switch (re-deploy with a new rule from strategies.html). Pin one. Post it in #trader-lab with a [ALUM] tag.
Agents Academy · 16 modules on tool-use, monitoring, and the production capstoneLimitless Trader Lab · complete
Congratulations, graduate.
One week ago you were clicking. Today you have a deployed bot that places real orders, logs every decision, and halts on a single file you control. Whatever happens to the bot from here, that’s the win.
One real live order ID on Limitless, placed by your deployed agent, not by a button click, with the place and cancel both logged in fills.ndjson.
A deployed bot on Railway with a proven kill-switch, you tripped it, watched it halt, deleted the file, and watched it resume.
A Claude + terminal + Trading / Strategy / Monitor skill stack you fully own, the same substrate you’ll extend in the Agents Academy and the Trader Lab Alumni community from here on.
From here: the Trader Lab Alumni role is permanent, #trader-lab stays open forever, the bonus pack lives on your laptop, and the three academies (PM, API, Agents) are yours to walk at your own pace. The cohort funnels into them, it doesn’t replace them.
Graduation call: tonight 7pm local on Google Meet · Trader Lab Alumni role granted within 1 hour of your post