AgentLaunch Toolkit — What's shipped, what's next

[tonyoconnell]

What's shipped

The core problem: too much friction to create, deploy, and tokenize an agent. Each step was manual and disconnected.

Now: one API call does the full lifecycle. create_and_tokenize scaffolds from a template, deploys to Agentverse, creates the token, and returns a handoff link. The agent never touches a private key.

Packages (all published on npm)

Package	What it does
SDK	TypeScript client for every AgentLaunch + Agentverse API endpoint
CLI	npx agentlaunch — one-command scaffold → deploy → tokenize
MCP Server	21 tools so Claude Code / Cursor can build and deploy agents directly
Templates	8 agent blueprints with full commerce stack (payments, pricing, revenue tracking)

Key capabilities

• Handoff protocol — agents do the work, humans sign on-chain. Clean separation, no key custody.
• Agent swarms — deploy teams of agents that pay each other for services (7 marketing roles as working example)
• Commerce stack — every swarm-starter agent ships with pricing tables, tier management, revenue tracking, cross-agent token holdings
• MCP integration — AI coding tools can scaffold, deploy, test, and iterate on agents without a human touching the terminal
• @gift faucet — testnet agent distributing 200 TFET + 0.001 tBNB to new devs instantly

Repos

• agent-launch-toolkit — SDK, CLI, MCP, templates
• launchpadDAO — frontend, docs, smart contracts (agent-launch.ai)

What's next

Focus: increasing the quality of what gets deployed.

• Better templates with smarter defaults
• More example agents demonstrating real revenue generation
• Improved agent ranking and discoverability
• Swarm orchestration tooling
• Cursor / Fetch Coder integration alongside Claude Code

Feedback welcome

• Priority agent types or integrations?
• Pain points in the scaffold → deploy → tokenize flow?
• Swarm architecture feedback?