Quest 1

Anatomy of a 1NC YOLO agent

Each agent is a single Node process that cycles once a minute, scans 650+ coins, enters on momentum, and self-manages exits.

Lesson

What a YOLO agent IS

A single PM2-managed Node process on the server. Each has:

Own state file (/tmp/agent-<id>-state.json) — positions, realized P&L, last scan
Own /status endpoint (localhost:PORT/status) — what the dashboard polls
Own force-trade endpoint (POST /force-trade) — manual sell
Own PM2 identity (pm2 list shows agent-<id>)

Inside each cycle (60s):

1. Bootstrap or reuse the 659-token Kraken USD universe 2. Fetch Kraken balance (ZUSD only — the cash you can actually spend) 3. Refresh BTC 1h EMA(20) — the regime gate 4. Check held positions for exit triggers 5. If none held AND BTC regime is bullish: scan, filter, pick, claim, buy 6. Save state, log telemetry, sleep 60s

Every 10s (between cycles): the ticker refresh updates the current_price of held positions without waiting for a full cycle.

The 10-stage entry pipeline

Already covered in the watchdog card's "YOLO v2 Active Rules" panel. Memorize these, they're what keep you out of BMB/CHEX-style traps.

Your task (quiz)

Answer the structural questions.

Quiz

Q1. Why does each agent have its OWN state file at /tmp/agent-<id>-state.json?

Legacy — the original agents all wrote to one fileFor isolation — a bug in one agent's state can't corrupt another's. Also each agent's positions, realized P&L, and track record are distinct.Because PM2 requires itTo avoid GitHub commits

Q2. The agent ticker refresh runs every 10 seconds. What does it update?

It re-scans the full 659-coin universeIt updates Kraken balance and places ordersIt fetches CURRENT prices for symbols the agent is HOLDING, so /status returns fresh mark-to-market and the equity chart moves intra-cycleIt restarts the PM2 process if cycle is frozen

1NC Platform Mastery