Pragmabase is in public beta — Try the playground free →
DocumentationPragmabase
v0.9 — Public Beta

Build agents that
think, remember, and pay.

SDK guides, runtime docs, memory architecture, x402 payment flows, and the Reliability Lab. Everything you need to deploy production-grade autonomous agents.

pragmabase SDK

// Install

npm install @pragmabase/sdk

// Initialize agent

import { Pragmabase } from '@pragmabase/sdk'

const pb = new Pragmabase({
  apiKey: process.env.PRAGMABASE_KEY,
  network: 'testnet', // or 'mainnet'
})

// Deploy with memory + wallet

const agent = await pb.agent.create({
  name: 'roku',
  memory: { enabled: true, team: 'research' },
  wallet: { budget: '10 XLM', guardrail: 'soroban' },
})

// agent.id → 0x4f2a...
console.log(await agent.run('Analyze market trends'))
All sections6 modules

Getting Started

Install the SDK, connect your Stellar wallet, and deploy your first autonomous agent in minutes.

Agent Runtime

Deep dive into the execution environment, lifecycle hooks, guardrail policies, and sandboxed simulation.

Memory Layer

Persistent vector memory powered by Qdrant. Store, retrieve, and evolve agent knowledge across runs.

Payments (x402)

x402 pay-per-action protocol on Stellar. Agents buy data, sell outputs, and run micro-economies.

Reliability Lab

Simulate failure scenarios before they have real consequences. Test, break, validate, then deploy.

API Reference

Complete REST API reference with request/response schemas, webhooks, and SDK code examples.

Architecture

How Pragmabase fits
into your stack.

Pragmabase sits between your agent logic and the real world. It wraps each agent run in a sandboxed environment, manages wallet state, routes memory reads/writes to Qdrant, and enforces Soroban-based guardrails before any funds move.

01Your agent logic calls the Pragmabase SDK
02Runtime injects memory context from Qdrant
03Soroban guardrail validates the action budget
04x402 payment executes on Stellar network
05Memory updated, audit trail recorded