What is the Schedly MCP server?

The Schedly MCP server (protocol version 2025-03-26) exposes 72 authenticated tools, 14 resources, and 15 built-in prompts that give AI agents complete access to your Schedly account — bookings, availability, customers, event types, payments, analytics, team, workflows, and more. Any MCP-compatible AI client (Claude, Cursor, Windsurf, ChatGPT, VS Code, n8n) can use natural language to read and write your Schedly data in real time.

Is the MCP server hosted or do I need to install anything?

Fully hosted — nothing to install. The primary endpoint is app.schedly.io/api/mcp (Streamable HTTP, MCP spec 2025-03-26). A legacy SSE endpoint at /api/mcp/sse is kept for older clients. Service discovery is available at /.well-known/mcp. No Docker, no npm install, no local process.

Which AI clients are supported?

Claude.ai and Claude Desktop use OAuth auto-discovery — add the URL, log in, done. Cursor also auto-discovers OAuth. Windsurf, VS Code (1.99+), Cline, Continue, ChatGPT Desktop, n8n, LangChain, CrewAI, AutoGen, and any MCP SDK client work with Bearer token auth. If your client speaks MCP 2024-11-05 or later, Schedly works with it.

What can I actually do with natural language?

Every action available in the Schedly dashboard. Examples: 'Create a 30-minute intro call for tomorrow at 3pm', 'Show me all no-shows from last month', 'Block every Thursday afternoon for the next 8 weeks', 'What was my revenue in Q2?', 'Cancel all bookings for July 4th', 'Who on my team has availability Monday morning?', 'List customers who haven't booked in 90 days'. The 72 tools cover the full scheduling lifecycle.

Are MCP tool calls secure?

Yes. OAuth calls are scoped to the authorised user's account. API key calls are scoped to the key owner's account. No tool can access another account's data. All calls appear in your audit log at app.schedly.io/settings/api-logs with timestamp, tool name, and parameters. Rate limits: 1,000 req/min on Pro, 5,000 req/min on Teams.

Why does the MCP server show as disconnected?

Common causes: (1) Wrong URL — the correct endpoint is app.schedly.io/api/mcp (note /api/). (2) Invalid or expired API key — regenerate at app.schedly.io/settings/developer/api-keys. (3) JSON syntax error in your config file — validate at jsonlint.com. (4) OAuth session expired — open your client's MCP settings and re-authorise. Still stuck? Email support@schedly.io with your client name, version, and the error message.

Can I use the MCP server in production agents?

Yes. The Schedly MCP server is production-ready, backed by the same 99.9% uptime SLA as the platform. Teams run it in n8n booking automations, LangChain intake agents, Slack bots that manage appointments, and internal copilots. The 72 tools, 14 resources, and 15 built-in prompts give agents everything they need for fully autonomous scheduling workflows.

MCP Server Live

Connect AI agents to Schedly

72 tools · 14 resources · 15 prompts. Full scheduling system access. Works with Claude, Cursor, ChatGPT, Windsurf — any MCP-compatible AI client.

Setup takes 60 seconds. No infrastructure to manage.

Open MCP Dashboard View Setup Guide ↓

Works with every MCP client

Claude

ChatGPT

Cursor

Windsurf

VS Code

n8n

Quick Start

Up and running in 3 steps

No server setup. No Docker. No install. OAuth for Claude & Cursor — API key for everything else.

Get your API key (or use OAuth)

Claude.ai, Claude Desktop, and Cursor auto-connect via OAuth — just paste the MCP URL and log in. For all other clients (Windsurf, VS Code, ChatGPT, n8n): go to Schedly → Settings → Developer → API Keys → Create key. Keys start with cal_.

Open API Key Settings →

Add Schedly to your MCP client

Pick your client in the config panel below and paste the snippet into the right file. OAuth clients need no API key — just the URL. Bearer token clients need your cal_... key in the Authorization header.

Start scheduling in natural language

Restart your client and start a new conversation. Try: "What does my schedule look like next week?" or "Create a 30-minute intro call for tomorrow at 3pm."

See example prompts →

Client Setup

Config for every MCP client

Copy-paste ready. Works with all major AI coding and chat clients.

# 1. Go to claude.ai → your profile → Settings
# 2. Click "Integrations" → "Add Integration"
# 3. Paste the MCP URL below and click Connect:

https://app.schedly.io/api/mcp

# 4. Claude auto-discovers OAuth via:
#    /.well-known/oauth-authorization-server
# 5. Log in to Schedly and authorise access.
#
# No API key needed — OAuth handles auth.
# Schedly appears in every claude.ai conversation.

ℹ️

Claude.ai uses OAuth 2.0 + PKCE — no API key to paste. The pre-registered client ID (anthropic-claude) is used automatically.

Example Prompts

What you can say to your AI agent

Every example below is a real tool call — not a simulation.

Bookings

›

"Book a 45-minute coaching call with Sarah Chen for this Friday at 2pm"

›

"Reschedule my 10am meeting tomorrow to 3pm"

›

"Cancel all bookings for the Photography Intro event type on July 4th"

›

"Show me everyone booked this week sorted by appointment time"

Availability

›

"What's my availability next Tuesday?"

›

"Block every Thursday from 12pm to 2pm for the next 8 weeks"

›

"Add a 30-minute buffer after all appointments in the Consultation schedule"

›

"Set me as unavailable December 24th through January 2nd"

Analytics

›

"What was my no-show rate last month by appointment type?"

›

"Show me revenue for Q2 2026 broken down by event type"

›

"Which day of the week has the most last-minute cancellations?"

›

"How many new customers booked in the last 30 days?"

Team & Customers

›

"Who on my team has availability for a new client on Monday morning?"

›

"List all customers who haven't booked in the last 90 days"

›

"Create a customer record for Alex Johnson, alex@example.com"

›

"Show me the full booking history for the customer Sarah Chen"

Tool Reference

72 tools · 14 resources · 15 prompts

Every tool is authenticated, scoped to your account, and production-ready. Click any category to see the full tool list.

Deep Dive

What the Schedly MCP server actually changes — and why it matters

The Model Context Protocol (MCP) is an open standard that lets AI systems call tools in authenticated external services using structured function calls. When your AI agent is connected to Schedly via MCP, it doesn't just answer questions about scheduling — it can actually create bookings, modify availability, retrieve customer records, and generate analytics reports. In real time. With your real data. The difference between an AI that talks about scheduling and an AI that does scheduling.

Before MCP, integrating AI with scheduling required building a custom API layer, managing OAuth tokens, and writing glue code to translate between AI outputs and API inputs. For most teams, that was a multi-week project. The Schedly MCP server collapses that to a config file and a restart. The 72 tools, 14 resources, and 15 built-in prompts are pre-built, pre-authenticated, and described with the rich schema MCP clients use to understand capabilities — including transport discovery at /.well-known/mcp and OAuth at /.well-known/oauth-authorization-server.

"Your AI agent gets the same capabilities as a logged-in service manager — every tool is authenticated, scoped to your account, and production-ready."

Building production agents with Schedly MCP

The most powerful use cases go beyond ad-hoc queries. Teams are embedding the Schedly MCP server into production workflows: n8n automations that create bookings when a CRM deal reaches "Proposal Sent", LangChain agents that handle the full intake-to-booking journey for new leads, Slack bots that let staff check and update schedules without leaving the channel, and internal copilots built on Claude or GPT-4 that answer questions like "Who's our busiest provider this month?" by actually querying live data rather than a stale report.

The architecture is intentionally simple. Your API key (or OAuth token) authenticates every call. All requests are logged. Rate limits are generous (1,000 req/min on Pro, 5,000 on Teams). The MCP server is fully hosted — no local process, no npx, no Docker. The primary transport is Streamable HTTP at app.schedly.io/api/mcp, with a legacy SSE endpoint at /api/mcp/sse kept for older clients.

Security model and what AI agents can and can't do

Every tool call is scoped to your Schedly account. An AI agent using your API key cannot access any other account's data. Delete operations (cancellations, customer removals, webhook deletions) require the AI to call the specific tool with the specific resource ID — there's no "delete everything" tool and no batch destructive operations. All calls appear in your Schedly audit log under Settings → API Logs with timestamp, tool name, and parameters — giving you a complete record of every action your AI agents have taken.

For enterprise deployments, Schedly's Teams plan supports multiple API keys with named labels — one per agent, one per environment. This lets you rotate keys per agent independently and trace activity by agent in the audit log.

FAQ

Frequently asked questions

MCP Server Live — Ready to connect

Give your AI agents
scheduling superpowers.

72 tools · 14 resources · 15 prompts. OAuth for Claude & Cursor. API key for everything else. Start free — connect your first agent in 60 seconds.

Open MCP Dashboard Start Free — No Credit Card