How do I use beds24-mcp?

1. Click on "Install Server". 2. Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state. 3. In the chat, type @ followed by the MCP server name and your instructions, e.g., "@beds24-mcp How does pricing propagate in Beds24?" That's it! The server will respond to your query, and you can continue using it as needed. Here is a step-by-step guide with screenshots.

beds24-mcp

by jverneuer

Overview Schema Related Servers Score Discussions

TypeScript

Local

beds24-mcp

Three independent npm packages for the Beds24 API, plus an MCP server that composes them:

Package	npm name	What it does
SDK	`beds24-sdk-client`	Typed V2 HTTP client + OpenAPI schema validation + typed domain ops (bookings, pricing, availability, channels, webhooks, inventory, accounts, properties, …).
Knowledge	`beds24-knowledge`	Markdown knowledge corpus, vector+FTS indexing, local embedding, hybrid (RRF) search.
Server	`beds24-mcp-server`	MCP server + CLI that wires the two packages into tools, prompts, and resources for an LLM.

All three packages are published independently to npm (Changesets, access: public). The root of this repo is a Bun workspace and is not itself published.

Architecture

The SDK and knowledge packages have no dependency on each other and no MCP or HTTP/embedding framework coupling. They are meant to be imported from other tooling (Inngest functions, Dagster assets, CLIs) without dragging in the server. The server is a thin shell that turns them into MCP tools.

┌──────────────────────────────┐
│        beds24-mcp-server     │  CLI (beds24-mcp-server index|status|serve|setup)
│   tools · prompts · resources│  MCP server (@modelcontextprotocol/sdk + zod)
└──────────────┬───────────────┘
       ┌───────┴────────┐
       ▼                ▼
┌─────────────┐  ┌──────────────┐
│beds24-sdk   │  │beds24-       │
│  -client    │  │  knowledge   │
│client · ops │  │indexer ·     │
│schema ·     │  │search ·      │
│validate     │  │embed         │
└─────────────┘  └──────────────┘
      │                 │
      ▼                 ▼
 apiV2.yaml        knowledge/*.md
 (bundled)         (bundled)

The server exposes the same surfaces in three groups, in the order an LLM should use them:

Phase	Surface	Why
Understand	`beds24_search*`, `beds24_howto`, `beds24_status`, facts resource	Read the cited docs before writing any code.
Validate	`beds24_schema`, `beds24_validate`, endpoints resource	Inspect the schema, then check payloads before spending credits.
Operate	`beds24_booking_`, `beds24_price_`, `beds24_availability_`, `beds24_inventory_`, `beds24_property_`, `beds24_account_`, `beds24_channel_`, `beds24_webhook_`	Typed V2 API calls against the live system.

Operational tools build a per-request Beds24Client from the caller's auth block (refreshToken | inviteCode | token) — the API key / prop key legacy flow is intentionally unsupported.

Related MCP server: schema-org-mcp

Quickstart

1. MCP route (end-user, AI harness)

npm install -g beds24-mcp-server   # publishes the beds24-mcp-server command
beds24-mcp-server setup            # detects Claude Code / Cursor / Windsurf / VS Code, writes their MCP config

Restart your harness — the tools appear in every session. setup is idempotent (it only touches the beds24 entry) and supports --dry-run, --harness <name>, --skip-index.

2. SDK route (typed V2 calls from your own code)

import { Beds24Client, BookingOps } from "@jverneuer/beds24-sdk-client";

const client = new Beds24Client({ refreshToken: process.env.BEDS24_REFRESH_TOKEN });
const { data, credits } = await new BookingOps(client).get({ arrivalFrom: "2026-09-01" });
console.log(`${credits.remaining} credits left`);

Every METHOD /path in apiV2.yaml is reachable via client.request("METHOD /path", body) with inferred types from the generated schemas. Request bodies are validated client-side before sending (fail fast, save a credit). See packages/sdk/README.md for the full ops list.

3. Knowledge route (search / index pipeline)

import { buildIndex, search } from "@jverneuer/beds24-knowledge";

// One-time (~30s) — builds the .beds24/index.db vector store.
await buildIndex({ knowledgeDir: "./knowledge" });

const hits = await search("how does pricing propagate to channels?", 5);
// hits: SearchHit[] — { text, sourceFile, headingPath, lines, bucket, docUrl, score }

Embedding runs locally via @huggingface/transformers (Xenova/all-MiniLM-L6-v2). No API keys.

Tools

All tools live in the server package. Each operational tool additionally takes an auth block:

{ refreshToken?: string; inviteCode?: string; token?: string; baseUrl?: string }

Provide one of refreshToken (preferred), inviteCode, or token. baseUrl defaults to https://www.beds24.com/api/v2.

Search & knowledge

Tool	Summary
`beds24_search`	Hybrid (vector + FTS) search — current apiv2 + general docs only.
`beds24_search_all`	Same, across ALL buckets (including deprecated apiv1).
`beds24_search_in_bucket`	Search one bucket (`apiv2` / `general` / `apiv1` / `deprecated`).
`beds24_howto`	End-to-end: search + schema + summarized steps for a task.
`beds24_status`	Index + corpus status (chunks, buckets, endpoints, facts files).

Schema & validation

Tool	Summary
`beds24_schema`	Resolve request/response schema for `METHOD /path` → flat field list.
`beds24_validate`	Validate a draft payload against the schema → structured errors.

Bookings

Tool	Summary
`beds24_booking_get`	GET /bookings (filter by status, dates, property/room).
`beds24_booking_create`	POST /bookings (typed `newBooking` shape).
`beds24_booking_cancel`	Cancel by id (cancellations, never deletes).
`beds24_booking_message_list`	GET /bookings/messages.
`beds24_booking_message_send`	POST /bookings/messages (OTA only).

Pricing

Tool	Summary
`beds24_price_set_daily`	POST /inventory/rooms/calendar (per-day prices).
`beds24_price_get_calendar`	GET /inventory/rooms/calendar.
`beds24_price_set_fixed`	POST /inventory/fixedPrices (date-range prices).

Availability

Tool	Summary
`beds24_availability_get`	GET /inventory/rooms/availability.

Inventory, properties, accounts

Tool	Summary
`beds24_inventory_offers`	GET /inventory/rooms/offers (arrival + departure + numAdults).
`beds24_property_list`	GET /properties (expand rooms, pictures, offers, …).
`beds24_account_list`	GET /accounts.

Channels & webhooks

Tool	Summary
`beds24_channel_settings_get`	GET /channels/settings.
`beds24_channel_settings_configure`	POST /channels/settings.
`beds24_webhook_register`	POST the webhook payload shape your URL receives.

Prompts

Prompt	Summary
`beds24_prompt_create_booking`	Walks through creating a booking (search → schema → validate → create).
`beds24_prompt_set_daily_prices`	Walks through setting daily prices.
`beds24_prompt_register_webhook`	Walks through the webhook payload shape.

Resources

URI	Contents
`beds24://facts/{path}`	One raw markdown facts file from the knowledge base.
`beds24://endpoints`	All V2 request/response endpoints as JSON.

Install from source

bun install      # install workspace deps
bun run index    # build the vector index from knowledge/*.md (one-time, ~30s)
bun run setup    # same harness auto-config as the global command
bun test         # run the full test suite
bun run typecheck

The server auto-indexes on startup if .beds24/index.db is missing, so bun run index is optional. bun run reindex forces a rebuild.

Source layout

.
├── packages/
│   ├── sdk/                       # beds24-sdk-client
│   │   ├── apiV2.yaml             # V2 OpenAPI spec (source of truth for the SDK)
│   │   ├── src/
│   │   │   ├── client.ts          # HTTP + V2 auth (inviteCode / refreshToken / token)
│   │   │   ├── api-types.ts       # generated-type helpers (EndpointKey, OpOf, …)
│   │   │   ├── ops/               # domain workflows (booking, pricing, channels, …)
│   │   │   ├── schema/            # spec resolution + ajv validation
│   │   │   └── generated/types.d.ts
│   │   ├── tests/                 # integration-style tests (mock fetch, recording client)
│   │   └── package.json
│   ├── knowledge/                 # beds24-knowledge
│   │   ├── knowledge/             # cited markdown facts + api-v2/, system-logic/, …
│   │   ├── src/
│   │   │   ├── db.ts              # libsql + sqlite-vec + FTS5
│   │   │   ├── embed.ts           # local Xenova/all-MiniLM-L6-v2 embedding
│   │   │   ├── indexer.ts         # walk → chunk → embed → store
│   │   │   ├── search.ts          # hybrid search (FTS + vector + RRF)
│   │   │   ├── markdown/          # heading-aware chunker + frontmatter parser
│   │   │   └── paths.ts
│   │   └── package.json
│   └── server/                    # beds24-mcp-server
│       ├── src/
│       │   ├── server.ts          # MCP tools / prompts / resources registration
│       │   ├── beds24.ts          # composition root (Beds24 facade over sdk + knowledge)
│       │   ├── cli.ts             # index | status | serve | setup
│       │   ├── setup.ts           # harness auto-config (Claude / Cursor / Windsurf / VS Code)
│       │   └── integration*.ts
│       └── package.json
├── .changeset/                    # Changesets (per-package independent versioning)
├── .github/workflows/
│   ├── ci.yml                     # typecheck + test on every push/PR
│   └── release.yml                # Version Packages PR → publish to npm on merge
├── scripts/
│   ├── fetch-spec.ts              # mirror apiV2.yaml from beds24.com
│   └── generate-types.ts          # generate types.d.ts from apiV2.yaml
├── CONTRACT.md                    # frozen cross-package interface contract
├── TEST-HARNESS.md                # test conventions every subagent follows
├── TYPESCRIPT-RULES.md            # codebase-wide TS discipline
└── package.json                   # Bun workspace root (private)

Publishing

Each package carries its own version and is released independently via Changesets. To propose a bump:

bun changeset        # interactive — pick package(s) + bump type

That writes a .changeset/*.md like:

---
"beds24-sdk-client": minor
---

Add InvoicingOps and InventoryOps

The release GitHub Actions workflow opens (or updates) a Version Packages PR that consumes pending changesets, bumps the affected packages, and writes changelog entries. When that PR lands on main (no pending changesets left), the same action runs changeset publish, which publishes only the packages whose version advanced. The server's workspace:* deps on its siblings are rewritten to concrete versions at publish time; updateInternalDependencies: "patch" in .changeset/config.json makes a dep bump flow a patch to dependents.

Required secret: NPM_TOKEN (Automation type). See .changeset/README.md for details.

Configure your harness (MCP)

All harnesses speak the same JSON shape; only the config file location differs.

The shared block (command / args). Prefer the global command when installed via npm — it survives reinstalls and doesn't hard-code a checkout path:

{ "command": "beds24-mcp-server", "args": ["serve"] }

Falling back to a source checkout (replace /ABS/PATH/to/beds24-mcp):

{ "command": "bun", "args": ["run", "/ABS/PATH/to/beds24-mcp/packages/server/src/server.ts"] }

Run beds24-mcp-server setup (or bun run setup) to write these files automatically — no hand-editing needed.

Claude Code

~/.claude/.mcp.json (user) or project .mcp.json. Uses the mcpServers key. Restart after editing.

Cursor

~/.cursor/mcp.json. Uses the mcpServers key. Open Settings → MCP; restart servers if it's not green.

Windsurf

~/.codeium/windsurf/mcp_config.json. Uses the mcpServers key. Reload the window.

VS Code (Copilot)

Project .vscode/mcp.json. Uses the servers key (note: not mcpServers). Restart after editing.

Troubleshooting

"bun not found" — launch the harness from a terminal (so it inherits $PATH), or add the bun dir to the config's env.PATH.
No tools after restart — the server auto-indexes on first run and logs to stderr. Open the harness's MCP panel and look for [beds24] MCP server connected on stdio..
Stale facts — after updating knowledge/, run bun run reindex (or delete .beds24/ and restart; it rebuilds).

This server cannot be installed

license - permissive license

Not graded

quality - not tested

maintenance

How are these scores calculated?

Maintenance

–Maintainers

–Response time

2dRelease cycle

6Releases (12mo)

Commit activity

Resources

Need Help?

Related Servers

Unclaimed servers have limited discoverability.

Looking for Admin?

If you are the server author, to access and configure the admin panel.

Related MCP Servers

mcp-openapi-schema
RAG Systems Documentation Access
hannesj
A
license
Not graded
quality
F
maintenance
A MCP server that exposes OpenAPI schema information to LLMs like Claude. This server allows an LLM to explore and understand large OpenAPI schemas through a set of specialized tools, without needing to load the whole schema into the context
Updated a year ago (2025-03-13 13:50 UTC)
113
50
MIT
schema-org-mcp
Documentation Access Developer Tools
Theycallmeholla
A
license
Not graded
quality
A
maintenance
MCP server enabling AI assistants to explore schema.org types, generate JSON-LD examples, validate structured data, and navigate the complete ontology with fuzzy matching and caching.
Updated a month ago (2026-07-13 10:20 UTC)
20
19
MIT
swagger-openapi-mcp
Documentation Access Code Analysis
joyboy-yy
A
license
Not graded
quality
C
maintenance
MCP server for querying Swagger/OpenAPI metadata efficiently from AI tools, enabling fast API discovery, search, and schema inspection.
Updated a month ago (2026-07-21 03:49 UTC)
409
MIT
cloudbeds-mcp-server
Real Estate Travel & Transportation
ronaldoribeirosm
A
license
A
quality
C
maintenance
MCP server that exposes Cloudbeds occupancy, check-ins/check-outs, and reservations as natural language tools, with read-only access and secure credential storage.
Updated 17 days ago (2026-08-04 12:59 UTC)
1
MIT

View all related MCP servers

Related MCP Connectors

1stay
Hotel booking MCP server. Search, book, and manage reservations across 250K+ properties worldwide.
Swagger MCP
MCP server for AI access to Swagger by SmartBear.
VeryChic Hotel Deals
Unofficial read-only MCP server for VeryChic hotel offers

View all MCP Connectors

Latest Blog Posts

Who's Calling? MCP Hosts Are an Identity Blind Spot (And the Spec Knows It)
By Om-Shree-0709 on July 25, 2026.
mcp
Agent Identity
OAuth 2.1
Your AI Chatbot Just Exposed Your CEO's Salary to an Intern
By Om-Shree-0709 on July 2, 2026.
Agent Identity
MCP Security
OAuth Delegation
Why MCP Servers Need Execution Sandboxing (And Why Your Current Stack Isn't Enough)
By Om-Shree-0709 on June 30, 2026.
Agentic Ai
Prompt Injection
WebAssembly

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/jverneuer/beds24-mcp'

If you have feedback or need assistance with the MCP directory API, please join our Discord server

beds24-mcp

Architecture

Quickstart

1. MCP route (end-user, AI harness)

2. SDK route (typed V2 calls from your own code)

3. Knowledge route (search / index pipeline)

Tools

Search & knowledge

Schema & validation

Bookings

Pricing

Availability

Inventory, properties, accounts

Channels & webhooks

Prompts

Resources

Install from source

Source layout

Publishing

Configure your harness (MCP)

Claude Code

Cursor

Windsurf

VS Code (Copilot)

Troubleshooting

Maintenance

Resources

Looking for Admin?

Related MCP Servers

mcp-openapi-schema

schema-org-mcp

swagger-openapi-mcp

cloudbeds-mcp-server

Related MCP Connectors

Latest Blog Posts

MCP directory API