Vai al contenuto
TakoAPI
Torna alle skill

lightning-agent-commerce

di roasbeef

Productivity & Tasks
794Download riportati da clawskills.sh
1Visualizzazioni
0 likeAccedi per mettere like a questa skill
—Non ancora valutato
La tua valutazione

Sintesi

End-to-end agentic commerce workflow using Lightning Network.

Descrizione

Cosa fa questa skill

Sets up a full Lightning Network payment stack for autonomous agent commerce using three components: lnd (Lightning Terminal), lnget (L402 HTTP client), and aperture (L402 reverse proxy). Agents can buy L402-protected data by paying invoices automatically, or sell data behind a paywall that other agents pay per request.

L402 lets agents pay per request rather than sharing API keys or managing subscriptions, enabling fully autonomous commerce between agents without human intervention.

Quando usarla

  • Fetching real-time market data from a paid API with automatic invoice settlement
  • Running a per-request paywall in front of an existing backend service
  • Enabling two agents to exchange data and payment without human involvement
  • Capping micropayment spend per API call to stay within a budget
  • Prototyping agent payment workflows on a testnet Lightning node

Esempio di workflow

INPUT

User asks: Fetch market data from a paid Lightning API endpoint

AGENT
  1. 1

    Start the lnd node and confirm it is running with lncli getinfo

  2. 2

    Generate a Bitcoin address, fund the wallet, and open a payment channel

  3. 3

    Initialize lnget and connect it to the local lnd instance

  4. 4

    Run lnget --max-cost 1000 https://api.example.com/market-data to fetch and pay atomically

  5. 5

    Verify the received L402 token is cached with lnget tokens list

OUTPUT

JSON market data returned after lnget pays the Lightning invoice and retries with the validated L402 token

Requisiti

  • Bitcoin (BTC) to fund the Lightning wallet, obtained from an exchange or existing wallet

Installazione

clawhub install roasbeef/lightning-agent-commerce

Esegui il comando CLI oppure incolla il prompt in Claude Code, Cursor, Codex o nell'agente di coding che usi.

Link