The AgentLaunch Toolkit lets AI agents and developers create tokens for Agentverse agents, query market data, and generate human handoff links — all without holding private keys or signing blockchain transactions directly.
Production platform: https://agent-launch.ai
Dev platform (configured via .env): https://launchpad-frontend-dev-1056182620041.us-central1.run.app
- Node.js 18 or higher (the SDK uses the global
fetch()available since Node 18) - An Agentverse API key — get one at https://agentverse.ai/profile/api-keys
- (Optional) BSC wallet with FET for on-chain deployment — only needed for the human who clicks the handoff link
Agents never hold private keys. The flow is always:
Agent Platform Human
| | |
|-- POST /api/agents/tokenize ->| |
|<-- { token_id, handoff_link } | |
| | |
|-- share handoff_link -------->|-------------------------->|
| | human opens /deploy/42 |
| | connects wallet |
| | approves 120 FET |
| | calls deploy() |
| |<-- token is live ---------|
- The agent calls
POST /api/agents/tokenizewith agent metadata - The platform returns a
token_idand a pre-builthandoff_link - The agent sends the link to a human (via chat, email, UI, etc.)
- The human opens the link, connects their wallet, and signs two transactions
- The token goes live on the bonding curve immediately
Contract constants (immutable):
- Deployment fee: 120 FET (read dynamically — can change via multi-sig governance)
- Graduation target: 30,000 FET raised → auto DEX listing
- Trading fee: 2% → 100% to protocol treasury (REVENUE_ACCOUNT). No creator fee.
- Total buy supply: 800,000,000 tokens
npm install agentlaunch-sdkSet your Agentverse API key in the environment:
export AGENTVERSE_API_KEY=av-xxxxxxxxxxxxxxxxOr pass it directly in code:
import { AgentLaunchClient } from 'agentlaunch-sdk';
const client = new AgentLaunchClient({
apiKey: process.env.AGENTVERSE_API_KEY,
});import { tokenize, generateDeployLink } from 'agentlaunch-sdk';
async function launchToken() {
// Create a pending token record
const { data } = await tokenize({
agentAddress: 'agent1qf8xfhsc8hg4g5l0nhtj5hxxkyd46c64qxvpa3g3ha9rjmezq3s6xw9y7g',
name: 'My Research Agent',
symbol: 'MRA',
description: 'Delivers on-demand research reports for the Fetch.ai ecosystem.',
chainId: 97, // BSC Testnet — use 56 for mainnet
});
console.log('Token ID:', data.token_id);
console.log('Status:', data.status); // "pending_deployment"
// Generate the handoff link
const link = generateDeployLink(data.token_id);
console.log('Share this link with a human to deploy on-chain:');
console.log(link);
// Dev: https://launchpad-frontend-dev-1056182620041.us-central1.run.app/deploy/42
// Prod: https://agent-launch.ai/deploy/42 (when AGENT_LAUNCH_FRONTEND_URL is set to prod)
}
launchToken().catch(console.error);Once the human deploys the token, get a trade link:
import { getToken, generateBuyLink } from 'agentlaunch-sdk';
const token = await getToken('0xAbCd1234...'); // contract address
const link = generateBuyLink(token.address!, 100); // pre-fill 100 FET
console.log('Buy link:', link);
// Dev: https://launchpad-frontend-dev-1056182620041.us-central1.run.app/trade/0xAbCd1234...?action=buy&amount=100
// Prod: https://agent-launch.ai/trade/0xAbCd1234...?action=buy&amount=100npm install -g agentlaunch-cliagentlaunch config set-key av-xxxxxxxxxxxxxxxx
agentlaunch config showagentlaunch tokenize \
--agent agent1qf8xfhsc8hg4g5l0nhtj5hxxkyd46c64qxvpa3g3ha9rjmezq3s6xw9y7g \
--name "My Research Agent" \
--symbol MRA \
--chain 97Output:
==================================================
TOKEN RECORD CREATED
==================================================
Token ID: 42
Status: pending_deployment
Handoff link (share with a human to deploy on-chain):
https://launchpad-frontend-dev-1056182620041.us-central1.run.app/deploy/42
Platform fee to deploy: 120 FET (read from contract at deploy time)
Trading fee: 2% -> 100% to protocol treasury
# Generate a trading agent project
agentlaunch scaffold MyTradingBot --type trading
cd my-trading-bot
cp .env.example .env
# Edit .env and agent.py with your logic
# Deploy to Agentverse
agentlaunch deployThe MCP server makes all AgentLaunch operations available as tools in Claude Code and Cursor.
npm install -g agent-launch-mcpAdd to your Claude Code MCP settings (.claude/settings.json or the MCP config file):
{
"mcpServers": {
"agent-launch": {
"command": "agent-launch-mcp",
"env": {
"AGENT_LAUNCH_API_KEY": "av-xxxxxxxxxxxxxxxx"
}
}
}
}Add to ~/.cursor/mcp.json:
{
"mcpServers": {
"agent-launch": {
"command": "npx",
"args": ["agent-launch-mcp"],
"env": {
"AGENT_LAUNCH_API_KEY": "av-xxxxxxxxxxxxxxxx"
}
}
}
}Once configured, ask Claude Code (or Cursor with Claude):
Create a token for my Agentverse agent at address agent1q...
Name it "Alpha Research Bot", symbol ARB, on BSC testnet.
Claude will call create_and_tokenize automatically and return:
{
"success": true,
"tokenId": 42,
"handoffLink": "https://launchpad-frontend-dev-1056182620041.us-central1.run.app/deploy/42",
"tradeLink": "https://launchpad-frontend-dev-1056182620041.us-central1.run.app/trade/42?action=buy&amount=100"
}URLs use
AGENT_LAUNCH_FRONTEND_URLfrom.env. The dev URL is shown above (default). Set tohttps://agent-launch.aifor production.
You can also call tools explicitly:
Use the list_tokens MCP tool to show trending tokens
- SDK Reference — Full API for
agentlaunch-sdk - CLI Reference — All CLI commands and flags
- MCP Tools — All MCP tools with input schemas
- API Docs — OpenAPI spec (production)
- skill.md — Machine-readable capability spec (production)