Loxtep

Agentic-first data platform for small business

Prefer to build hands-on? It's five commands

Most people just tell Loxtep what agent they want. But if you'd rather build the data foundation yourself in code, the whole lifecycle is five commands: init → attach → generate → test → deploy. Here's exactly what each step looks like in your terminal.

This is a guided, read-only walkthrough — the terminal output below is illustrative and nothing here runs commands or connects to your account. To run it for real, install the authenticated CLI: npm i -g @loxtep/sdk.

loxtep — getting startedstep 01
loxtep init --template shopify-orders
✓ Scaffolded .loxtep/project.json ✓ Created domains/ connectors/ workflows/ data-products/ ✓ Wrote AGENTS.md and .loxtep/skills/shopify-orders.yaml Next: loxtep attach · Getting Started → docs.loxtep.io/start
1loxtep init

Scaffolds an agent-ready project from a template — config, directory tree, an AGENTS.md for your coding agent, and a default skill scoped to just this template's resources.

loxtep — getting startedstep 02
loxtep attach --instance prod
✓ Linked project to managed instance lx_prod_8f3a ✓ Wrote instance_id + api_url to .loxtep/project.json
2loxtep attach

Connects the project to a shared, managed, or self-hosted instance using the same mechanism as the Platform UI, then records the resolved instance.

loxtep — getting startedstep 03
loxtep generate
Generating typed workspace context… data products 12 connectors 5 domains 3 queues 8 ✓ Wrote .loxtep/generated/index.ts
3loxtep generate

Projects your live data products, connectors, domains, and queues into typed constants your agent and SDK reference by name — no guessing identifiers.

loxtep — getting startedstep 04
loxtep test orders-enricher --event ./events/order-created.json
▶ Running orders-enricher locally ─── Action Trace ─────────────────────────────────── ✓ [1] handler.start → orders-enricher [succeeded] ✓ [2] queues.read → orders_raw [succeeded] ✓ [3] dataProducts.write → orders_enriched [succeeded] ✓ [4] handler.complete → orders-enricher [succeeded] ──────────────────────────────────────────────────── ✓ 1 event processed, 0 errors
4loxtep test

Executes a workflow handler locally with a sample event and prints the full action trace. Operations listed in requireApproval prompt for confirmation before executing.

loxtep — getting startedstep 05
loxtep deploy
Compiling 3 modules… ✓ Validating referenced resources… ✓ Updating orders-enricher (in place) ✓ Creating refund-watcher ✓ ✓ Deploy queued · track: loxtep activity list
5loxtep deploy

Compiles every module, checks that referenced resources exist before touching anything, then drives in-place updates and creates through the workflow engine.

What test and deploy actually run

Workflows are authored in code with the typed SDK. Triggers and sinks reference the same generated workspace constants, so a typo is a compile error — not a 2 a.m. incident.

Operations you mark as guarded pause for human approval before they touch live data, and every run records an action trace you can review with loxtep activity list.

workflows/orders-enricher.tstypescript
import { defineDataWorkflow, on } from '@loxtep/sdk'
import { workspace } from './.loxtep/generated'

// Author a data workflow in code, against typed workspace constants.
export default defineDataWorkflow({
  name: 'orders-enricher',
  triggers: [on.queueEvent(workspace.queues.orders_raw)],
  async handler(ctx, event) {
    await ctx.toolbox.dataProducts.write(
      workspace.dataProducts.orders_enriched,
      event,
    )
  },
})

// 'loxtep deploy' compiles this module into the workflow graph and ships it.

Stop wrestling with data plumbing. Start launching agents that work.

Join the waitlist for early access. Tell Loxtep what agent you want, connect your tools, and ship an agent your business can actually rely on — without building a data engineering team.

Enterprise-grade reliability, permissions, and audit trails — without the enterprise team.

Free to join — no credit card, no commitment