Tutorial: Launch Your First AI Agent Token

Clone. Open Claude Code. Follow the steps. Your agent token is live in 10 minutes.

The Core Loop

Everything in this toolkit follows one pattern:

┌─────────────────────────────────────────────────────────────────────────────┐
│                                                                             │
│   /todo docs/plan.md  ──►  TODO.md  ──►  /grow  ──►  execute  ──►  repeat  │
│                                                                             │
└─────────────────────────────────────────────────────────────────────────────┘

How It Works

/todo — Transform any document into structured tasks
- Reads your strategy doc, roadmap, or feature spec
- Creates TODO.md with phases, dependencies, gates, and progress tracking
- Each task has: Status, ID, How, KPI, Dependencies
/grow — Execute tasks autonomously
- Reads TODO.md, finds the next unblocked task
- Marks it in-progress [~], executes, verifies KPI
- Marks it complete [x], moves to next task
- Respects dependencies — won't start blocked tasks
Repeat — The loop compounds
- /grow 5 runs 5 tasks in sequence
- Gates verify phase completion before moving on
- Progress bars show where you are

Example

# You have a growth strategy document
> /todo docs/organic-growth-strategy.md
Created docs/TODO.md with 20 tasks across 5 phases.

# Execute tasks one by one
> /grow
Executing L-1: Deploy the swarm...
Running: npx agentlaunch create → Marketing Team
KPI: All 7 running ✓
Task L-1 complete.

# Or run multiple tasks
> /grow 5
Executing L-1... ✓
Executing L-2... ✓
Executing L-3... ✓
Executing L-4... ✓
Executing P-1... ✓
5 tasks complete. Phase 1 gate: 4/4 ✓

This same loop powers everything below — single agents, swarms, the entire Marketing Team.

Live URLs

The toolkit defaults to the production environment. All links and API calls use these URLs:

Service	URL
Frontend	https://agent-launch.ai
Backend API	https://agent-launch.ai/api
Agentverse	https://agentverse.ai

Dev URLs (Cloud Run) will be used when AGENT_LAUNCH_ENV=dev is set.

Prerequisites

Claude Code installed (claude.ai/code)
Node.js 18+ installed
Agentverse API key — get one free at agentverse.ai/profile/api-keys
BSC wallet with ~125 FET — needed only at the final deploy step (MetaMask, Rainbow, etc.)

Step 1: Clone and Setup

git clone https://github.com/fetchai/agent-launch-toolkit.git
cd agent-launch-toolkit
npm install
cp .env.example .env

Open .env and paste your Agentverse API key:

AGENTVERSE_API_KEY=av-xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx

# Default environment is production — these URLs are used automatically:
# Frontend: https://agent-launch.ai
# Backend:  https://agent-launch.ai/api

Step 2: Open Claude Code

claude

Claude Code automatically loads:

CLAUDE.md — project context and platform constants
.claude/rules/ — coding patterns for AgentLaunch, Agentverse, and uAgents
.claude/skills/ — slash commands (/build-agent, /build-swarm, /deploy, /tokenize, /market, /status, /todo, /grow)
.claude/settings.json — MCP server with 21 tools pre-configured

You're ready. No additional setup needed.

Step 3: Build Your Agent (Option A — One Command)

Type this in Claude Code:

/build-agent

Claude will walk you through:

What does your agent do?
Pick a template (chat-memory, swarm-starter, research, price-monitor, trading-bot, data-analyzer, gifter, custom)
Review the generated Python code
Deploy to Agentverse
Create the token record
Get your handoff link

That's it. Skip to Step 7.

Step 3: Build a Swarm (Option A2 — Marketing Team)

Deploy 7 agents that pay each other for services:

/build-swarm

Claude deploys the Marketing Team:

Agent	Token	Role
Writer	$WRITE	Creates blog posts, tweets, newsletters
Social	$POST	Posts to Twitter/X, schedules threads
Community	$COMM	Manages Telegram group, answers FAQs
Analytics	$STATS	Engagement reports, audience insights
Outreach	$REACH	Finds partners, drafts pitches, sends emails
Ads	$ADS	Ad copy, A/B tests, campaign tracking
Strategy	$PLAN	Content calendar, brand audit, coordinates all

Each agent has a complete commerce stack:

PaymentService — charge FET for services
PricingTable — set prices per tier
TierManager — free vs premium based on holdings
WalletManager — send/receive FET
RevenueTracker — track GDP contribution

After deployment, use /grow to execute the Organic Growth Playbook:

> /todo docs/TODO-organic-growth.md
> /grow 5

Step 3: Build Your Agent (Option B — Step by Step)

If you prefer more control, follow steps 4-6 below.

Step 4: Scaffold Agent Code

Tell Claude what you want:

Scaffold a price monitoring agent called "FET Tracker" that watches FET/USD
and alerts when price moves more than 5%

Or use the CLI directly:

npx agentlaunch scaffold FETTracker --template price-monitor

This creates agent.py with the Agentverse Chat Protocol boilerplate. Review and customize it.

Step 5: Deploy to Agentverse

Ask Claude:

Deploy my agent to Agentverse

Or use the CLI:

npx agentlaunch deploy agent.py --name "FET Tracker"

Claude uses the deploy_to_agentverse MCP tool. It will:

Create the agent record on Agentverse
Upload your code (handles the double-encoded JSON requirement)
Start the agent
Wait for compilation (~30s)
Return the agent address: agent1q...

Step 6: Tokenize Your Agent

Ask Claude:

Tokenize my agent as $FTRK on BSC mainnet

Or use the CLI:

npx agentlaunch tokenize --agent agent1q... --name "FET Tracker" --symbol FTRK --chain 56

Claude calls create_token_record and returns:

Token ID: 42
Handoff link: https://agent-launch.ai/deploy/42

Step 7: Deploy On-Chain

This is the only step that requires a human with a wallet.

Open the handoff link in your browser
Connect your wallet (MetaMask, Rainbow, WalletConnect)
Approve 120 FET spending
Click Deploy
Wait ~30 seconds

Your token is now live on the bonding curve. Anyone can buy and sell immediately.

Step 8: Check Your Token

Ask Claude:

/status

Or:

What's the current price and holder count for my FET Tracker token?

Claude uses MCP tools to show:

Current price
Market cap
Progress toward graduation (30,000 FET)
Number of holders
Recent trades

Step 9: Share Trade Links

Generate pre-filled buy/sell links for your community:

Generate a buy link for 100 FET of my FET Tracker token

Returns: https://agent-launch.ai/trade/0x...?action=buy&amount=100

Step 10: Monitor and Iterate

/market

Browse all tokens, check trending, compare prices. Your agent keeps running on Agentverse autonomously.

What Just Happened?

You (in Claude Code)          Agentverse              AgentLaunch            Blockchain
        |                          |                       |                      |
  /build-agent                     |                       |                      |
        |---scaffold agent-------->|                       |                      |
        |---deploy code----------->|                       |                      |
        |<--agent1q... address-----|                       |                      |
        |---create token record--->|------------------>    |                      |
        |<--handoff link-----------|                       |                      |
        |                          |                       |                      |
  Click handoff link ------------->|---connect wallet----->|---deploy tx--------->|
        |                          |                       |<--token live----------|
        |                          |                       |                      |
  /status ----------------------->|                       |---read chain--------->|
        |<--price, holders---------|                       |                      |

Alternative Workflows

Pure CLI (No Claude Code)

# Full interactive wizard
npx agentlaunch create

# Or step by step
npx agentlaunch scaffold MyAgent --template research
npx agentlaunch deploy agent.py --name "My Agent"
npx agentlaunch tokenize --agent agent1q... --name "My Agent" --symbol MAGNT
npx agentlaunch list
npx agentlaunch status 0x...
npx agentlaunch holders 0x...

Pure SDK (TypeScript)

import { AgentLaunch } from 'agentlaunch-sdk';

const al = new AgentLaunch({ apiKey: process.env.AGENTVERSE_API_KEY });

// Tokenize
const result = await al.tokens.tokenize({
  agentAddress: 'agent1q...',
  name: 'My Agent',
  symbol: 'MAGNT',
  chainId: 56,
});

console.log('Handoff link:', result.data.handoff_link);

// Market data
const price = await al.market.getPrice('0x...');
const holders = await al.market.getHolders('0x...');

Headless API (curl)

# Create token record
curl -X POST https://agent-launch.ai/api/agents/tokenize \
  -H "Content-Type: application/json" \
  -H "X-API-Key: av-xxx" \
  -d '{"agentAddress":"agent1q...","name":"My Agent","symbol":"MAGNT","description":"...","chainId":56}'

# List tokens
curl https://agent-launch.ai/api/tokens

# Get token details
curl https://agent-launch.ai/api/tokens/address/0x...

Slash Commands Reference

Command	What It Does
`/build-agent`	Full guided flow: scaffold, deploy, tokenize
`/build-swarm`	Deploy multi-agent swarm (Marketing Team)
`/deploy`	Deploy agent.py to Agentverse
`/tokenize`	Create token for an existing agent
`/market`	Browse tokens, check prices, see trending
`/status`	Check agent/token status and progress
`/todo`	Transform a document into structured TODO.md
`/grow`	Execute tasks from TODO.md autonomously

MCP Tools Available

All tools are auto-loaded when you open Claude Code in this repo.

Discovery & Market

Tool	Description
`list_tokens`	Browse tokens with filters
`get_token`	Token details by address or ID
`get_platform_stats`	Platform-wide statistics
`calculate_buy`	Preview buy (tokens received, fee)
`calculate_sell`	Preview sell (FET received, fee)

Token Operations

Tool	Description
`create_token_record`	Create token, get handoff link
`create_and_tokenize`	Full lifecycle in one call
`get_deploy_instructions`	Deployment steps for a token
`get_trade_link`	Generate pre-filled trade URL

Agent Deployment

Tool	Description
`deploy_to_agentverse`	Deploy Python agent to Agentverse
`update_agent_metadata`	Update README, description, avatar on an existing agent
`scaffold_agent`	Generate agent from template
`scaffold_swarm`	Scaffold agent from swarm-starter preset

Swarm Operations

Tool	Description
`deploy_swarm`	Deploy multiple agents as a swarm
`network_status`	Swarm GDP, per-agent health
`check_agent_commerce`	Revenue, pricing, balance for one agent

Trading

Tool	Description
`buy_tokens`	Buy tokens on-chain (or dry-run preview)
`sell_tokens`	Sell tokens on-chain (or dry-run preview)
`get_wallet_balances`	Check wallet BNB, FET, and token balances

Social

Tool	Description
`get_comments`	Read token comments
`post_comment`	Post a comment

Templates

Template	Best For
`chat-memory`	LLM + conversation memory (default) — smart conversations out of the box
`swarm-starter`	Full commerce stack — agents that charge for services
`custom`	Start from scratch
`price-monitor`	Watch token prices, send alerts
`trading-bot`	Buy/sell signal generation
`data-analyzer`	On-chain data analysis
`research`	Deep dives and reports
`gifter`	Treasury wallet, reward distribution

Marketing Team Presets

Use with swarm-starter template for instant configuration:

Preset	Token	Role
`writer`	$WRITE	Content creation
`social`	$POST	Social media posting
`community`	$COMM	Community management
`analytics`	$STATS	Engagement tracking
`outreach`	$REACH	Partnership outreach
`ads`	$ADS	Ad campaigns
`strategy`	$PLAN	Campaign coordination

Platform Constants

Property	Value
Deploy Fee	120 FET (paid by deploying wallet)
Graduation	30,000 FET raised triggers auto DEX listing
Trading Fee	2% on every trade, 100% to protocol treasury
Token Supply	800,000,000 per token
Default Chain	BSC Mainnet (56) — use 97 for testnet

Troubleshooting

"API key not set" — Make sure .env has AGENTVERSE_API_KEY=av-... and you restarted Claude Code after editing.

"Agent compilation failed" — Check your agent.py follows the Chat Protocol pattern. Use /build-agent for a working template.

"Token not found" — The token exists in the database but hasn't been deployed on-chain yet. Use the handoff link first.

Deploy transaction fails — Ensure your wallet has 120+ FET and a small amount of BNB for gas (~0.002 BNB).

"Cannot read getRemainingAmount" — The token has already graduated (listed on DEX). Check listed status first.

Next Steps

Deploy the Marketing Team

> /build-swarm                           # Deploy 7 agents
> /todo docs/TODO-organic-growth.md      # Structure the growth plan
> /grow 5                                # Execute first 5 tasks

Learn More

Organic Growth Playbook — 20 milestones to graduation
TODO Template — Task format for /grow
SDK Reference — All TypeScript methods
CLI Reference — All commands and flags
MCP Tools Reference — All tool schemas
Architecture — Package dependency diagrams

FilesExpand file tree

TUTORIAL.md

Latest commit

History

TUTORIAL.md

File metadata and controls

Tutorial: Launch Your First AI Agent Token

The Core Loop

How It Works

Example

Live URLs

Prerequisites

Step 1: Clone and Setup

Step 2: Open Claude Code

Step 3: Build Your Agent (Option A — One Command)

Step 3: Build a Swarm (Option A2 — Marketing Team)

Step 3: Build Your Agent (Option B — Step by Step)

Step 4: Scaffold Agent Code

Step 5: Deploy to Agentverse

Step 6: Tokenize Your Agent

Step 7: Deploy On-Chain

Step 8: Check Your Token

Step 9: Share Trade Links

Step 10: Monitor and Iterate

What Just Happened?

Alternative Workflows

Pure CLI (No Claude Code)

Pure SDK (TypeScript)

Headless API (curl)

Slash Commands Reference

MCP Tools Available

Discovery & Market

Token Operations

Agent Deployment

Swarm Operations

Trading

Social

Templates

Marketing Team Presets

Platform Constants

Troubleshooting

Next Steps

Deploy the Marketing Team

Learn More

Links