LivePair Bots

Production-ready bot templates that put the LivePair AI image & video generation API inside Discord, Telegram, X/Twitter, and n8n.

Every template is zero-dependency — Node 20+ built-ins only, no discord.js, no bot frameworks. Clone, set env vars, run. Billing is per-result against prepaid lp_ credits: no subscription, no monthly tier. Renders auto-delete after 48h; nothing enters a public feed.

Contents

• Requirements
• Quick start
• Bots
• Configuration
• Deployment
• n8n workflows
• Architecture
• Error handling
• Related projects

Requirements

• Node.js ≥ 20 (uses built-in fetch, crypto, http — nothing else)
• A LivePair API key — create one at livepairai.com/settings
• Platform credentials for the bot you're running (see Configuration)

Quick start

git clone https://github.com/livepairai/livepair-bots.git
cd livepair-bots
cp .env.example .env    # fill in your keys

node telegram-bot.mjs   # or discord-bot.mjs / twitter-bot.mjs

Bots

File	Platform	Interface	Billing
discord-bot.mjs	Discord	/imagine + /video slash commands	owner's lp_ key
telegram-bot.mjs	Telegram	long-polling, any message → media	owner's lp_ key
twitter-bot.mjs	X/Twitter	one-shot CLI poster (cron/CI friendly)	owner's lp_ key

Discord

Slash commands via the interactions webhook — no gateway connection, no discord.js. Request signatures are verified with Node's built-in ed25519 crypto.

# 1. Create an app at discord.com/developers → copy PUBLIC KEY + APP ID
# 2. Register commands once (GUILD_ID for instant testing; global takes ~1h)
APP_ID=… BOT_TOKEN=… node register-commands.mjs
# 3. Run the webhook server
PUBLIC_KEY=… APP_ID=… LP_KEY=lp_… node discord-bot.mjs
# 4. Set the app's Interactions Endpoint URL → https://<host>:8787/interactions

Built-in behavior:

• Model picker — optional model option resolved against the live catalog (/v1/agent/models, cached 5 min). Defaults: qwen-image-3 for images, wan-3.0-t2v for video; override with IMAGE_MODEL / VIDEO_MODEL.
• Private-line gating — model ids containing -private answer only in bot DMs or NSFW-flagged guild channels, using Discord's own channel flag. Public channels get an ephemeral nudge.
• Quota-free by design — every generation bills your key; add your own rate limiting (env, KV, per-user auth) if you need it.

Telegram

Long-polls getUpdates; any message text becomes a generation prompt and the bot replies with the rendered media.

BOT_TOKEN=123:abc… LP_KEY=lp_… node telegram-bot.mjs

BOT_TOKEN comes from @BotFather. Set IMAGE_MODEL for any catalog id; VIDEO=1 switches replies to sendVideo for video models.

X/Twitter

One-shot generate-and-post — designed for cron, CI jobs, and schedulers. OAuth 1.0a signing is implemented inline with built-in crypto.

node twitter-bot.mjs "a cat astronaut, cinematic" --text "daily drop"
node twitter-bot.mjs "slow wave crash" --video --model wan-3.0-t2v

X credentials come from your own app at developer.x.com (Read+Write). The free tier posts tweets and uploads media — sufficient for a poster bot. Async video uploads are polled via the media-upload STATUS command.

Configuration

Variable	Used by	Source
LP_KEY	all	livepairai.com/settings — prepaid credits
BOT_TOKEN	telegram, register-commands	@BotFather / Discord dev portal
PUBLIC_KEY, APP_ID	discord	Discord developer portal
X_API_KEY/X_API_SECRET/X_ACCESS_TOKEN/X_ACCESS_SECRET	twitter	developer.x.com app
IMAGE_MODEL / VIDEO_MODEL	discord, telegram	any id from GET /v1/agent/models
LIVEPAIR_BASE	all	API base override (default https://livepairai.com)
PORT	discord	webhook listen port (default 8787)

See .env.example for a ready-to-copy template.

Deployment

Render (one click):

render.yaml ships a free-plan web service for the Discord bot — fill PUBLIC_KEY, APP_ID, LP_KEY in the dashboard, then point the app's Interactions Endpoint URL at the deployed /interactions path.

Docker / any VPS:

docker build -t livepair-bots . && docker run -p 8787:8787 --env-file .env livepair-bots
# Railway: railway up  ·  Fly: fly deploy  ·  bare metal: node discord-bot.mjs

n8n workflows

Import via Workflows → Import from File:

• n8n-telegram-workflow.json — Telegram trigger → generate → poll → sendPhoto.
• n8n-x-workflow.json — webhook → LivePair Generate Media (community node) → Post to X. Requires @livepairai/n8n-nodes-livepair
  • X OAuth2 credentials.

Architecture

┌────────────┐   webhook/polling   ┌──────────────┐   HTTPS    ┌─────────────────┐
│  Platform   │ ──────────────────→ │  bot script   │ ─────────→ │ livepairai.com  │
│ (Discord/… │                     │  (this repo)  │            │ /v1/agent/*     │
└────────────┘                     │               │            └─────────────────┘
                                   │  lib/livepair.mjs            submit → jobId
                                   │  · generate()  submit+poll   poll  → media URL
                                   │  · listModels() catalog      (48h TTL)
                                   │  · LivePairError typed errs
                                   └──────────────┘

One API contract backs everything: POST /v1/agent/generate → jobId → GET /v1/agent/jobs/{id} → media URL. Two payment rails: lp_ prepaid key, or x402 (agents pay per call in USDC on Base — see x402-payment.md in the examples repo). Agent-facing config snippets for Hermes/OpenClaw and any OpenAI-compatible framework live in hermes-openclaw.md.

OpenAI-compatible surfaces

Already inside an OpenAI-shaped client? Point it at https://livepairai.com/v1/agent — /chat/completions, /images/generations, and /videos (+ GET /videos/{id}, /videos/{id}/content) all speak the stock spec shapes, and GET /models returns the OpenAI object:"list" shape so model pickers populate. Same lp_ key, same credits billing.

Error handling

The shared client throws LivePairError with an HTTP status:

Status	Meaning	Action
401/403	bad or missing lp_ key	fix credentials
402	insufficient credits (or an x402 quote)	top up at /settings
429	rate limited	back off
5xx	server error	safe to retry
—	job failed/timeout	prompt or provider issue; check error message

Failed generations don't settle — you're only billed for results.

Related projects

• livepair-examples — API cookbook (livepair.py client, mobile proxy, x402)
• livepair-cli — npm i -g livepair terminal client
• n8n-nodes-livepair — n8n community node
• ComfyUI-LivePair — ComfyUI nodes
• livepair-skills — agent SKILL.md · OpenAPI · MCP

License

MIT — see LICENSE. Contributions welcome.