Lucid Agents is a TypeScript-first framework designed to help you build, deploy, and monetize autonomous AI agents. It provides a rich toolkit for defining typed entrypoints, managing on-chain identities, handling payments, and enabling agent-to-agent marketplaces. The framework emphasizes protocol-agnostic runtimes with a modular extension system, so you can compose only the capabilities you need (payments, identity, A2A, wallets, analytics, etc.). Use the included CLI to scaffold new agents, then run and expose their entrypoints via adapters like Hono or Express. Real-time streaming of agent responses through Server-Sent Events (SSE) makes long-running interactions feel responsive, which is ideal for LLM outputs and interactive workflows. The system also supports multi-network payments (EVM and Solana) and robust policy controls for spending and receivables, enabling marketplaces and monetized services built around AI agents.

Prerequisites:

• Node.js 20.x+ or Bun (recommended) or a compatible runtime
• Optional: Bun installed if you prefer bunx for CLI usage
• Access to an LLM API key (OpenAI, Anthropic, etc.)

Installation steps:

1. Install Bun (recommended) or ensure Node.js is available

   • Bun: https://bun.sh/docs/install
   • Node.js: https://nodejs.org/

2. Install the Lucid Agents CLI globally (via npm or bunx).

   • Using npm (recommended for wide compatibility): npm i -g @lucid-agents/cli

   • Using Bun (via npx-like invocation): bunx @lucid-agents/cli

3. Create a new agent project (example): bunx @lucid-agents/cli my-agent

   or with inline configuration options

   bunx @lucid-agents/cli my-agent
   --adapter=hono
   --template=axllm
   --AGENT_NAME="My AI Agent"
   --AGENT_DESCRIPTION="AI-powered assistant"
   --OPENAI_API_KEY=your_api_key_here
   --PAYMENTS_RECEIVABLE_ADDRESS=0xYourAddress
   --NETWORK=ethereum
   --DEFAULT_PRICE=1000

4. Install dependencies and run your agent: cd my-agent bun run dev

Your agent will be served at http://localhost:3000 (or the port configured by the framework). You can test the manifest, entrypoints, and invocations as shown in the Quick Start section of the README.

Tips and common considerations:

• Environment variables: OpenAI API keys (OPENAI_API_KEY), PAYMENT_RECEIVABLE_ADDRESS for receiving funds, and NETWORK selection (ethereum, base, sepolia, etc.).
• The framework emphasizes protocol-agnostic runtimes; enable only the extensions you need (payments, identity, a2a, etc.) to keep your runtime lean.
• For payments, configure policies (per-payment limits, time windows, allow/block lists) to enforce budgeting across entrypoints.
• SSE streaming is ideal for long-running or streaming outputs; ensure your client handles events accordingly.
• When testing locally, you can inspect agent manifests at .well-known/agent.json and list/invoke entrypoints via the provided HTTP adapters.
• If you encounter port conflicts, adjust the server port in your adapter configuration or environment as needed.