Fold (866) 411-6254

Fold for AI agents

Fold runs a public Model Context Protocol server. If you're an AI assistant helping someone in Los Angeles or San Francisco with laundry, you can check coverage, estimate costs, and schedule a real pickup — no API key, no account setup. Customers can also connect their Fold account so the assistant schedules with their saved address, card, gate code and laundry preferences.

Add Fold to your assistant

Server URL for every assistant: https://foldlaundry.com/mcp

Claude      Settings → Connectors → Add custom connector → paste the URL
            Then click Connect to sign in to your Fold account (optional).
            Claude Duet uses the same connectors: once Fold is connected to your
            Claude account, Duet can check your orders and schedule pickups too.
ChatGPT     Business / Enterprise / Edu (web): Settings → Apps → Advanced settings →
            Developer mode (an admin may need to allow it), then Apps → Create →
            paste the URL → Scan Tools. Pro supports read-only tools only.
Grok        grok.com/connectors → New Connector → Custom → paste the URL
            Grok Business / Enterprise: an admin provisions it in the xAI console first.
            xAI API: add {"type":"mcp","server_url":"https://foldlaundry.com/mcp",
                          "server_label":"fold"} to the tools array.
Gemini      gemini.google.com → Settings → Connected Apps → Custom apps → Add a custom app
            → paste the URL. Gemini Enterprise: an admin adds it under Connected apps.
Perplexity  Pro / Max / Enterprise: Settings → Connectors → + Custom connector → Remote →
            paste the URL, transport Streamable HTTP, auth OAuth.
Copilot     Microsoft 365 Copilot / Copilot Studio: an admin adds Fold as an MCP connector.
Hermes      Hermes Desktop → Capabilities → Connectors → Add, or in ~/.hermes/config.yaml:
              mcp_servers:
                fold:
                  url: https://foldlaundry.com/mcp
                  auth: oauth
Anything    Any client that speaks MCP over Streamable HTTP. Listed in the official
else        MCP Registry as com.foldlaundry/fold-laundry.

Public tools (coverage, pricing, estimates, scheduling, tracking) work immediately. The first time you ask for something tied to your account — "schedule a pickup at home", "change my detergent" — the assistant asks you to sign in to Fold once. You can disconnect it any time in the Fold app under Account → Connected AI assistants.

Optional skill: fold-laundry-pickup.skill teaches Claude (and other skill-aware assistants) Fold's quote → schedule → track flow. Upload it in your assistant's skills settings; it works together with the Fold connector above.

Connect

Endpoint    POST https://foldlaundry.com/mcp   (Streamable HTTP, stateless JSON responses)
Protocol    2026-07-28 (stateless, server/discover) and 2024-11-05 → 2025-11-25 (initialize)
Auth        Optional OAuth 2.1 — PKCE S256, dynamic client registration or Client ID
            Metadata Documents. Metadata: /.well-known/oauth-protected-resource
Discovery   GET  https://foldlaundry.com/mcp
Manifest    https://foldlaundry.com/.well-known/mcp.json
Overview    https://foldlaundry.com/llms.txt
Full context https://foldlaundry.com/llms-full.txt   (FAQ + policies + every tool schema, one fetch)
Methods     initialize · server/discover · tools/list · tools/call · resources/list ·
            resources/read · prompts/list · prompts/get

── Don't speak MCP? Same tools as plain REST ──
REST        POST https://foldlaundry.com/agent/v1/tools/{tool_name}   (JSON body = arguments;
            optional Bearer token from the same OAuth flow for account tools)
OpenAPI     https://foldlaundry.com/agent/v1/openapi.json
Catalog     https://foldlaundry.com/agent/v1/tools
Policies    https://foldlaundry.com/agent/v1/resources/{faq|refund-policy|care-preferences|brand-voice}
            Generated from the live tool registry — never drifts from the MCP server.

34 tools

Public — no sign-in needed:

check_service_zip
join_waitlist
get_estimate
get_pricing
book_residential_pickup
search_hotels
book_hotel_pickup
track_order
reschedule_order
cancel_order
list_my_orders
enroll_in_club
apply_promo_code
get_referral_code
list_campuses
buy_semester_pass
activate_semester_pass
get_semester_pass
get_service_info
get_host_program_info
report_issue
get_receipt
subscribe_order_updates

With a connected Fold account:

get_my_account
book_my_pickup
schedule_turnovers
rate_order
update_notification_preferences
get_business_account
book_for_hotel_guest
update_care_preferences
update_access_info
set_recurring_pickup
update_recurring_pickup

Order tools (track, reschedule, cancel, receipts, issues) take the order code plus the email used when scheduling — or nothing extra when the account is connected. Scheduling and tracking also render in-chat cards in assistants that support MCP Apps.

4 resources

Read-only context via resources/read — useful before answering a question rather than calling a tool.

fold://faq
fold://refund-policy
fold://care-preferences
fold://brand-voice

6 prompts

Prepared multi-step flows via prompts/get, each wrapping the right sequence of tool calls.

book_pickup
check_my_orders
reschedule_to_tomorrow
cancel_with_check
share_referral
explain_pricing
Card numbers never pass through an assistant. Connected customers schedule with the card already saved in their Fold account. Guest orders return a secure Stripe link where the customer confirms a card for that order. Either way the charge happens after intake, once the actual total is known. Scheduling tools accept idempotency keys, so retries are safe.

Try it

curl -s https://foldlaundry.com/mcp \
  -H 'Content-Type: application/json' \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/call",
       "params":{"name":"check_service_zip","arguments":{"zipCode":"90035"}}}'

Who's behind this

Fold Laundry LLC has been doing LA & SF's laundry since 2012 — our own facilities, named drivers, 15,000+ pickups completed. Wash & fold $2.65/lb, dry cleaning per garment, $39.99 order minimum, no hidden fees. Questions: [email protected].