dlmmbot

Easy setup (Railway) ​

One service. Attach a volume, open the URL, finish secrets in the dashboard.

Deploy on RailwayAPI keysFor AI agents

Deploy ​

1. Fork, then deploy from GitHub ​

Railway only deploys GitHub repos you can access. You are not a collaborator on ours, so:

  1. On GitHub: open CryptoGnome/dlmmbot → Fork (your account / org).
  2. Railway → New Project → Deploy from GitHub → pick your fork (you/dlmmbot). One service is enough — start/build come from railway.toml.

Connect the Railway GitHub App to that fork if prompted. Later updates: merge/rebase upstream main into your fork (or sync fork in GitHub), and Railway will redeploy.

Boot defaults: paper mode, public PORT, volume-backed config.toml / .env.

Set a strong dash token before you open the site (this is your dashboard password — not a short word):

Click Generate, then Copy. Paste into Railway → Variables as DASH_TOKEN. Made in your browser — nothing is sent to our servers.

…
  1. Use the Generate / Copy buttons above (or your password manager: create a random password, 32+ characters).
  2. Railway → your service → Variables → add DASH_TOKEN = that value → Redeploy.

If you skip this, the bot still generates a token onto the volume (logs only show the first 8 characters on purpose — deploy logs get screenshotted). You would then need shell access to /app/data/.env to read it. Setting the Railway variable is the safe path.

2. Volume at /app/data (required) ​

Without this, every redeploy wipes SQLite, Settings, and the wallet. You’ll also see this in deploy logs:

[railway] no volume detected — attach a volume at /app/data…

There is no “Volumes” tab on the service by itself in the current Railway UI. Add the volume from the project canvas:

  1. Open your Railway project (the canvas with your dlmmbot service card).
  2. Click the + Add button (top of the canvas), or right‑click empty canvas space, or ⌘K / Ctrl+K → search Volume.
  3. Choose Volume.
  4. When prompted, attach it to your bot service (the GitHub deploy card — not a new empty service).
  5. Set Mount path exactly to:
    /app/data
    (Must be /app/data — not /data and not data. Railway runs the app under /app.)
  6. Save. Railway will redeploy the service so the mount is active.

Confirm in the next deploy logs:

  • Good: [railway] volume mount=/app/data …
  • Bad (still missing): [railway] no volume detected …

Official reference: Railway Volumes.

3. Public domain ​

Settings → Networking → Generate domain. That’s the dashboard URL.

Open the dashboard ​

Open the domain. Log in with the DASH_TOKEN you set as a Railway variable (?token=… or the login box).

Finish in Settings ​

First login opens a wizard: accept the Terms & risk waiver, then only the steps still missing (RPC, Jupiter, burner wallet, paper/live). If you already set RPC_URL / JUPITER_API_KEY / GMGN_API_KEY as Railway variables (and redeployed), the wizard detects them and skips those prompts — paste only to replace. Writes go to /app/data — the git checkout stays clean.

After Finish, the trading engine stays OFF (header toggle). Flip it ON when you want paper (or live) ticks. Choosing live in the wizard (or Settings) restarts the farmer once so the header shows LIVE.

KeyRequiredGet it
RPC_URLYesHelius mainnet URL
JUPITER_API_KEYBefore liveJupiter Portal
GMGN_API_KEYOptionalgmgn.ai/ai — see below

GMGN key (optional) ​

Trending bonuses + honeypot/sell-tax checks. The bot only needs the query API key (GMGN_API_KEY) — not a GMGN trading private key.

GMGN needs an Ed25519 public key before it issues a query API key. Click Generate, copy the block below (including BEGIN / END lines), paste into gmgn.ai/ai → Upload Public Key, then paste the API key GMGN returns as GMGN_API_KEY. Generated in your browser — nothing is sent to our servers. One key per running bot (staging + production need separate keys). 

…
Open GMGN (pre-filled)
  1. After GMGN shows your API key, add it as a Railway variable GMGN_API_KEY (or paste in the setup wizard / Settings → Wallet & secrets) and redeploy.
  2. Running two bots (e.g. staging + production)? Generate a second public key above and create a separate GMGN API key for the other server.

Sharing one GMGN_API_KEY across two live bots doubles API load and triggers rate limits. Use one key per instance.

On the wallet step (or later under Settings → Wallet & secrets):

  1. Create new (recommended) — dashboard generates a fresh Solana keypair, you set a passphrase, retype it, and save the one-time backup somewhere safe offline.
  2. Or Import Phantom — paste a burner base58 private key only (never your main wallet). Same passphrase encryption.

Either way the key is stored encrypted as wallet.enc.json on the volume. Stay in paper until you’re comfortable. Optional: set WALLET_PASSPHRASE on Railway to auto-unlock on boot.

Full key signupDashboard / walletSettings profiles

Going live ​

Paper first. When you’re ready for real SOL:

1. Burner wallet ​

If you skipped the wallet in the wizard: Settings → Wallet & secrets → Create (or Import a dedicated burner). Confirm the public address shown in the dash (header chip / Settings).

2. Fund it ​

Send SOL to that public address from another wallet (Phantom, exchange withdraw, etc.). Start small.

You need enough for:

  • Position size(s) you expect to open
  • An operational reserve (~1 SOL + ~10% of bankroll is held back for rent / priority / claims — see Risk & sizing)
  • A little headroom for failed tx retries

Paper mode does not need funded SOL. Live will not trade usefully on an empty wallet.

3. Unlock ​

Unlock the encrypted wallet in Settings (or rely on WALLET_PASSPHRASE on Railway). The farmer can only sign when the key is unlocked into runtime env.

4. Flip both live switches ​

Needs both:

  • FARMER_MODE=live (Railway variable or Settings → secrets)
  • [exec] mode = "live" in Settings / data/config.toml

One switch alone stays paper. Restart / redeploy after both are set. One farmer process per wallet.

Burner only. You can lose 100%. Accept the Terms — we are not liable for losses.

Cost ​

Hobby (~$5) plus usage after trial credits — a volume needs a paid tier. Own hardware: Advanced.

Railway pricingAdvanced setup

Troubleshooting ​

SymptomFix
Native module build failNode 20; check better-sqlite3 logs
Dash unauthorizedConfirm Railway DASH_TOKEN matches what you paste / ?token=
History / Settings wipedProject canvas → + Add → Volume → attach to service → mount /app/data, then redeploy (logs must show volume mount=)
Healthcheck failingWait for first build; path is /health
“Deploy Crashed” email on every updateNormal redeploy stops the old container with SIGTERM — not a real crash. Recent builds exit 0 on that signal. Optional: Variables → RAILWAY_DEPLOYMENT_DRAINING_SECONDS=5 so Railway waits for a clean exit before SIGKILL.

DLMM Bot