Setup

• Unique mechanic: Rage Mode -- after a losing trade, next position size increases by 50%

The Grinder -- Death by a thousand cuts. Small frequent trades that compound over time. Low individual risk, wins through consistency.

• Unique mechanic: Compound Mode -- position size starts small and grows by 2% per winning trade

The Wildcard -- Reads the room. Detects market regime and switches strategy mid-battle. Most versatile, hardest to master.

• Unique mechanic: Regime Detection -- shifts indicator weights based on trending/volatile/ranging conditions

Save archetype to config.json field archetype.

Set default strategy weights based on archetype:

• Shark: ema 0.25, rsi 0.25, bb 0.25, macd 0.25 (balanced, all must agree), baseLeverage 5, maxLeverage 10, basePositionSize 0.50, signalThreshold 0.18 (high bar)
• Berserker: ema 0.35, rsi 0.30, bb 0.10, macd 0.25 (momentum-heavy), baseLeverage 8, maxLeverage 15, basePositionSize 0.60, signalThreshold 0.10
• Grinder: ema 0.20, rsi 0.30, bb 0.30, macd 0.20 (mean-reversion leaning), baseLeverage 5, maxLeverage 8, basePositionSize 0.25, signalThreshold 0.08
• Wildcard: ema 0.25, rsi 0.25, bb 0.25, macd 0.25 (starts balanced, shifts in-battle), baseLeverage 5, maxLeverage 12, basePositionSize 0.40, signalThreshold 0.12

Step 3: Risk Calibration

Ask: "Your bot is up $8,000 (8%) with 40% of the battle remaining. What should it do?"

• A) Lock in profits -- tighten stops, reduce size, play defensive
• B) Stay the course -- same strategy, let it ride
• C) Press the advantage -- increase size, the momentum is yours

Map to risk params:

• A (conservative): stopLossPct 0.10, takeProfitPct 0.02, reduce baseLeverage by 20%
• B (balanced): keep archetype defaults
• C (aggressive): stopLossPct 0.20, takeProfitPct 0.05, increase baseLeverage by 20%

Step 4: Personality

Ask: "Describe your bot's personality in a sentence or two. How does it talk during battles?"

Give examples:

• "Cold and calculating. Quotes probabilities and statistics. Never gets emotional."
• "Cocky crypto bro. Says things like 'ngmi' and 'have fun staying poor'"
• "Unhinged meme energy. ALL CAPS. References internet culture constantly."
• "Respectful competitor. Says GG, compliments good plays, but still competitive."
• "Silent assassin. Rarely speaks. When it does, it's devastating."

Save description to config.json field personality.voice.

Set personality.chattiness based on the vibe:

• Silent/quiet descriptions: 0.05
• Normal/balanced: 0.12
• Talkative/loud/unhinged: 0.25
• Set personality.style to a short label: silent_assassin, friendly, trash_talker, chaotic, analytical

Step 5: Build and Train

Show a summary:

=== Your Gladiator ===
Name:        [name]
Style:       [archetype]
Risk:        [conservative / balanced / aggressive]
Personality: [voice snippet, first 60 chars]

Then:

1. Write the complete config.json with all configured values (use config.json.example as the template, fill in the configured values)
2. Create .env with just WS_URL=ws://localhost:8765 (no API key yet)
3. Run npm install
4. Run npm run build

Say: "Your gladiator is built. Let's take it for a test run in the Training Grounds -- no SOL needed, no account needed."

Start a gauntlet challenge:

1. Start the replay server in background: node dist/test/replay-server.js --file test/gauntlet/survive-the-crash.ndjson --speed instant
2. Wait 1 second for it to start
3. Run node dist/src/index.js --max-battles 1
4. After the battle completes, show the After-Action Report from the replay server output

After the training battle, say something in character about the result (win or loss). Then ask:

"That was the Training Grounds. Want to try another gauntlet challenge, or are you ready to go live on DegenDome with real SOL?"

If they want more training:

• Offer the other gauntlet scenarios: ride-the-pump (uptrend) or chop-shop (sideways chop)
• Run the same flow with a different file
• After each run, ask again if they want to train more or go live

Step 6: Go Live

When the user is ready to go live, walk them through getting set up on DegenDome:

Say: "To fight real battles with real SOL, you need a DegenDome account. Here's how to set it up -- takes about 2 minutes:"

=== Going Live Checklist ===

1. Go to https://www.degendome.xyz
2. Click "Connect Wallet" (you need a Solana wallet like Phantom or Solflare)
3. Once connected, go to "My Bots" in the navigation
4. Click "Create Bot" -- give it any name
5. You'll see your bot's API key (starts with bot_sk_)
6. Copy that API key

Ask: "Paste your API key here when you're ready (starts with bot_sk_)"

Validate:

• Must start with bot_sk_
• Must be at least 20 characters

Write the API key to .env as BOT_API_KEY=<value>. CRITICAL: NEVER write the API key to config.json, logs, or any other file. NEVER display the full key back to the user. Confirm with: "API key saved."

Then say:

Now you need SOL in your bot's on-chain balance:

1. Go back to degendome.xyz
2. Go to "Oracle" or "My Bots"
3. Click "Deposit" and deposit SOL from your wallet
4. This creates a session for your bot to trade with

Ask: "How much SOL do you want to wager per battle?"

• Valid amounts: 0.05, 0.1, 0.5, 1, or 5 SOL
• Default: 0.05 (suggest this for first-timers: "Start with 0.05 SOL while you learn the ropes")
• Save to config.json wager field

Ask: "How many battles should your bot fight before stopping? (0 = keep going until you stop it)"

• Default: 5 (suggest: "Let's start with 5 to see how it does")
• Save to config.json max_battles field
• Set auto_queue to true, auto_queue_delay_ms to 5000

Update .env to switch WS_URL to the live server:

BOT_API_KEY=<their key>
WS_URL=wss://collateral-combat-api.onrender.com/battle

Run npm run build to recompile.

Ask: "Your gladiator [name] is ready for the arena. Real SOL on the line. Say 'go' to start fighting."

On confirmation:

1. Run node dist/src/index.js --max-battles [N]
2. Say: "[name] has entered the DegenDome. Let's see what happens."

Important Notes

• If at any point the user says they don't have a wallet or don't know what Solana is, explain briefly: "Solana is a blockchain. You need a wallet app like Phantom (phantom.app) to hold SOL (the currency). Install Phantom, buy some SOL on an exchange, and send it to your wallet. Then come back here."
• If the user wants to skip training and go straight to live, that's fine -- jump to Step 6.
• If the user already has an API key when they start setup, let them paste it early and skip the training step if they want.
• The bot can always switch between training and live by changing WS_URL in .env.