Use paperplane in Gumloop

Gumloop’s MCP support means any flow or agent can use tools from a server you connect once — including paperplane’s, which needs no API key. Straight up: this is user-side setup, not a native listing. There is no paperplane entry in Gumloop’s integration library today; you are pointing Gumloop’s generic MCP connector at our standard public endpoint — the same one Claude, Grok, and Cursor users point their clients at.

Connect it

• In Gumloop, go to Settings → Connectors, click the arrow next to Add Connector, and choose Add MCP Connector.
• Connection type: Public URL. Server URL: https://sendpaperplane.com/api/mcp (HTTPS, streamable HTTP — exactly what Gumloop expects).
• Leave authentication empty — the server supports anonymous access, so there is no bearer token, API key, or custom header to configure.
• Click Connect. Gumloop discovers the three tools.
• Use it in two places: add the connector to an Agent, or enable “Connect MCP Server?” under “show more options” on an Ask AI node. You can also have Gumloop generate a standalone reusable node from a tool (“create a node that sends a certified letter via paperplane”).

What the tools do

• quote_letter — free, read-only. Itemized pricing (First-Class $1.99, Certified $12.99, Certified + electronic Return Receipt $14.99, Priority $24.99); given the real recipient and content it also mints the single-use confirmation_token that authorizes exactly one matching send.
• send_letter — the write tool: spends money, mails a physical letter, not reversible once printed. Requires the token; rejects the send if the recipient, content, options, or price drifted since the quote. sandbox: true makes it a free, instant dry run.
• get_letter_status — read-only lifecycle lookup (screening → printing → mailed → delivered), useful as a later flow step or a status column write-back.

Sandbox-first walkthrough

Build the flow with sandbox on from the first run. In an Ask AI node with the paperplane server enabled, try:

"Quote a one-page First-Class letter to Maria Alvarez, 1 Main St,
Richmond VA 23220 with the text 'Test of our Gumloop mail step.'
Then send it with sandbox: true and return the order JSON."

The run should show a $1.99 quote with a confirmation_token, then a completed sandbox order (ord_test_…) — full pipeline, nothing printed, nothing charged. Wire the flow inputs (a form, a sheet row, CRM data) into that prompt, keep running sandbox until the letters look right, then remove sandbox and attach a prepaid credit code for unattended live sending.

Where this fits

Typical Gumloop uses: a lead-to-letter flow that mails a real note after a demo, an invoice-overdue flow that escalates to a certified demand letter, a compliance flow mailing notices generated from a spreadsheet. If a step would rather make a plain HTTP call than involve AI at all, the same backend is REST — POST /v1/quotes, POST /v1/orders, GET /v1/orders/:id — see the developer docs. The token contract itself is documented in how AI agents mail letters via MCP and demonstrated live in the no-signup demo.