for developers · mcp

Ask the live Peppol network in plain language.

PeppolStatus runs a public remote MCP server: 29 read‑only tools that put the live Peppol network one question away from your agent — the access points and SMPs we monitor around the clock, the full directory of millions of participants, company‑register enrichment for Belgium and France, provider SLAs, certificate history, and country adoption. It mirrors our public /v1 API, so your agent reads the same data our dashboards do. No account needed to start — keyless works for everyone; an optional API key raises your rate limit.

Examples

What you can ask

Once it's connected, you ask in your own words and the client picks the tools. A few things it can answer from live data:

Is Acme NV registered on Peppol, and which access point routes its invoices? participant lookup → resolves the SMP and the provider behind it
Which access point lost the most participants this month, and who moved away? access-point churn → ranked leavers, with the participants named
How many French companies are reachable on Peppol now, and what share of the market is that? country adoption → live counts against SIRENE company-register facts
Has any access point I depend on had an outage or a certificate change in the last 24 hours? hosts, incidents & anomalies → verdicts, layer timings, and cert rotations
Show Belgium's Peppol adoption over the last year, month by month. adoption history → snapshots you can chart or diff
Which providers clear the 99.5% availability bar, and where does mine sit? provider SLA → monthly availability against the agreement bar

Setup

Add PeppolStatus to your client

Paste one URL. Every current MCP client speaks the same transport, so setup is a copy‑paste — pick yours below.

Claude keyless

In claude.ai, open Settings → Connectors, choose Add custom connector, and paste the endpoint URL:

Claude custom connector endpoint URL

https://mcp.peppolstatus.com

API keys — beta. Sending a key with Claude connectors relies on org‑admin static_headers, which is in beta and currently misfires (upstream #644 / #112). Use the keyless flow above; if you need the higher limit today, prefer Claude Code, Cursor or VS Code, which send headers natively.

Claude Code keyless

Add the server from your terminal with the Claude Code CLI:

Claude Code CLI commands to add the server

claude mcp add --transport http peppolstatus https://mcp.peppolstatus.com

# optional — raise your rate limit with an API key
claude mcp add --transport http peppolstatus https://mcp.peppolstatus.com \
  --header "x-api-key: pps_live_your_key"

ChatGPT anonymous only

On a paid plan, open Settings → Connectors (enable Developer mode if required), add a connector, and paste the endpoint URL:

ChatGPT connector endpoint URL

https://mcp.peppolstatus.com

No custom headers. ChatGPT connectors cannot send an API key, so ChatGPT always uses the keyless anonymous tier. Everything works — you just share the anonymous rate limit.

Cursor keyless

Add an entry to ~/.cursor/mcp.json (global) or .cursor/mcp.json in your project. The headers block is optional:

Cursor mcp.json configuration

{
  "mcpServers": {
    "peppolstatus": {
      "url": "https://mcp.peppolstatus.com",
      "headers": { "x-api-key": "pps_live_your_key" }
    }
  }
}

VS Code keyless

Add a servers entry to .vscode/mcp.json in your workspace (type http). Add a headers block with your key to raise the limit:

VS Code mcp.json configuration

{
  "servers": {
    "peppolstatus": {
      "type": "http",
      "url": "https://mcp.peppolstatus.com",
      "headers": { "x-api-key": "pps_live_your_key" }
    }
  }
}

How it works

One endpoint, read-only, honestly rate-limited

The server speaks the Streamable HTTP transport every current MCP client understands, at a single URL. Every tool is read‑only — there is no write path to the Peppol network here — so you can add it without a second thought. Each tool call is one request against the same public API, under the same 60 requests / 60 seconds limit.

Keyless callers share an anonymous per‑IP budget — plenty to try things out. When it gets tight, create a free API key at app.peppolstatus.com/keys; it gives your calls their own higher, attributed budget. Send it as Authorization: Bearer <key> or x-api-key: <key> — never in the URL — and the console prints the snippets above pre‑filled with your key. The tools are the same /v1 REST surface: same data, same shapes, wrapped for agents.

Audience

Who it's for

Engineers wiring up agents

Give an agent grounded Peppol facts instead of guesses — participant lookups, provider footprints, live network health — all read‑only and safe to hand to a model.

Analysts working in Claude or ChatGPT

Ask questions in the client you already use and get answers from live data: adoption trends, churn between providers, uptime and incident history — no SQL, no scraping.

Vendors embedding Peppol intelligence

Wire the same tools into your own product or internal copilot. Because it mirrors the /v1 API, what you prototype over MCP ports straight to REST when you need it.

Related resources

API reference (Scalar, interactive) The full /v1 REST surface the MCP tools mirror — browse and try every endpoint. https://api.peppolstatus.com/docs
API keys console Create a free key to raise your rate limit; copy pre-filled MCP snippets. https://app.peppolstatus.com/keys
Developer resources OpenAPI 3.1 spec, quickstart, auth guide, and agent discovery files. https://peppolstatus.com/developers/

Get started

Point your agent at the Peppol network

Paste the endpoint into your client and start asking — no key required. Need a higher limit or attributed usage? Grab a free API key. Built and run by the engineers at e‑invoice.be; questions go to hello@peppolstatus.com.

Get a free API key