Integration surfaces
🤖 Native AI-agent access · REST + MCP · Live USD pricing

Voltaxion for AI agents and B2B systems.

Voltaxion exposes live catalog access through JSON feeds, a read-only public API, an MCP server and discovery files. Phase 1 is anonymous read. Phase 2 adds customer-scoped Bearer keys and real quote locks.

Read-only in production Catalog and MCP reads are live. No Phase 1 write path is positioned as a confirmed stock allocation.
Operationally strict Respect rate limits, cache responsibly and treat quote responses as indicative until Phase 2 locks exist.
Three pillars

Three ways to integrate, one commercial surface.

Use direct feeds for ingestion, MCP for agent tooling, or discovery files for automatic capability detection.

REST and JSON

Static JSON for ingestion, API for controlled reads

Phase 1 exposes a large sample catalog and image map over HTTPS, plus a public read-only API on api.voltaxion.com.

  • /data/catalog_sample.json for SKU, price, stock, grade and routing metadata
  • /data/product_images.json for product-to-image lookup
  • https://api.voltaxion.com for public read-only catalog access
MCP server

Purpose-built tool layer for agents

The MCP surface narrows the catalog into four operator-safe tools: filtered search, category enumeration, SKU lookup and a Phase 1 pseudo quote request.

  • search_catalog for brand/category/model/condition/max-price filters
  • list_categories for category discovery and validation
  • get_sku for full detail by VTX SKU
  • request_quote for Phase 1 indicative quoting
Discovery files

Machine-readable entry points for tool-aware clients

Voltaxion publishes llms.txt and ai-plugin.json on the public domains so clients can discover scope and access boundaries without scraping the site.

  • https://voltaxion.com/llms.txt plus matching portal and admin variants
  • https://voltaxion.com/.well-known/ai-plugin.json plus matching portal and admin variants
  • Use discovery first when your client supports automatic capability mapping
Access model

Anonymous read now. Customer-scoped auth next.

The live posture is conservative: public discovery, public reads and no claim that a Phase 1 quote response equals a stock reservation.

Phase 1 live

Anonymous read surfaces

Today you can fetch sample data, browse discovery files and call MCP catalog tools without a customer-specific Bearer token.

  • Read-only JSON and API access is live
  • MCP catalog search and SKU detail are live
  • request_quote returns a Phase 1 pseudo quote, not a hold or reservation
Phase 2 staged

Bearer keys and real quote locks

Phase 2 introduces customer-bound API keys, per-customer rate ceilings and real quote locking backed by server-side state.

  • Authorization: Bearer vtx_live_... is the planned auth shape
  • Quote locks and stock holds are a roadmap item, not a Phase 1 promise
  • Use contact for early-access discussion
Endpoints

Surface map

Current integration touchpoints.

Live https://voltaxion.com/llms.txt Mirrored on portal and admin domains.
Live https://voltaxion.com/.well-known/ai-plugin.json Published on all three public-facing domains.
Live https://api.voltaxion.com Public read-only catalog API.
Live https://mcp.voltaxion.com/mcp MCP server exposing four catalog tools.
Live https://portal.voltaxion.com/data/catalog_sample.json Sample feed with 1200+ SKUs.
Pending https://portal.voltaxion.com/.well-known/openapi.json Until it is published, use llms.txt, ai-plugin.json and the sample catalog JSON.

Keep staging copies out of indexation until the OpenAPI file is live.

Rate limits

Operational limits are explicit.

Cache aggressively, back off on 429, and do not assume the HTML site, data feeds and admin endpoints share the same threshold.

Surface Path scope Limit Expected behavior
Catalog feeds /data/*.json 30 requests/minute per IP, burst 10 Best for cached pulls and scheduled syncs. Overages return 429.
Page endpoints HTML pages on voltaxion.com 120 requests/minute per IP Suitable for human browsing and lightweight agent discovery, not bulk harvesting.
Admin endpoints Admin app/API surfaces 600 requests/minute per IP, burst 200 Reserved for authenticated internal workflows and signup/onboarding flows.
Code examples

Start simple, then move up the stack.

Most clients begin with the catalog sample feed, then switch to MCP for structured agent workflows.

curl

Fetch the sample catalog feed

The sample feed is the fastest way to validate field coverage before deeper wiring.

curl -sS https://portal.voltaxion.com/data/catalog_sample.json \
  -H "Accept: application/json"
Python httpx

Read and inspect the JSON feed

Keep the contract plain: fetch, raise on transport failure, then inspect the response shape.

import httpx

url = "https://portal.voltaxion.com/data/catalog_sample.json"

with httpx.Client(timeout=20.0) as client:
    response = client.get(url, headers={"Accept": "application/json"})
    response.raise_for_status()
    payload = response.json()

print(type(payload))
print(payload.keys() if isinstance(payload, dict) else "list payload")
MCP

Search the catalog through tool invocation

Use the MCP surface when you want a constrained tool contract instead of raw file parsing.

{
  "tool": "search_catalog",
  "arguments": {
    "brand": "Apple",
    "category": "smartphone",
    "model": "iPhone 15",
    "condition": "Grade A",
    "max_price_usd": 700
  }
}
Schema and discovery

Use discovery files as the contract index.

The sample feed is large enough to show the working field shape: VTX SKU, brand, model, grade, stock, unit price in USD, route or hub metadata and image references. Treat llms.txt as the public map and ai-plugin.json as the current capability summary.

Usage policy

Allowed uses are narrow by design.

This is a B2B catalog surface for sourcing workflows and controlled automation.

Allowed

Good uses of the live Phase 1 surface

Workflows this page is meant to support today.

  • Read catalog sample data to evaluate assortment, grade mix and pricing posture.
  • Use MCP tools to search categories, inspect SKU detail and draft sourcing workflows.
  • Cache responsibly inside internal buyer tools, ERP pilots or procurement copilots.
  • Use pseudo quote responses for evaluation and internal workflow testing.
Not allowed

What not to assume or do

Current operational boundaries.

  • Do not treat a Phase 1 pseudo quote as a confirmed lock, stock hold or shipping commitment.
  • Do not bypass, spray or parallelize around the published rate limits.
  • Do not republish the raw feed as a public mirror or a free third-party catalog dump.
  • Do not represent unpublished or pending surfaces as production contracts.
Q3 2026 roadmap: Voltaxion plans to move from indicative quote flows to programmatic quote-and-lock with authenticated customer keys, per-customer controls and server-side stock holds. Until then, the correct framing is discovery + evaluation + read-only integration.

Need a deeper integration path?

Use the live public surfaces to evaluate data shape today. For authenticated access planning and early Phase 2 discussions, start with a commercial onboarding conversation.

Contact Voltaxion Apply for access →