tickadoo
Official# tickadoo MCP Server
`@tickadoo/mcp-server` is the local stdio entrypoint for tickadoo MCP. Since v2.0.0 it is a thin bridge to the canonical remote server at:
```text
https://mcp.tickadoo.com/mcp
```
## Meta Muse connector
Directory submission pack and the brief Muse should fetch live in [`connectors/muse/`](connectors/muse/). Stable brief:
https://raw.githubusercontent.com/tickadoo/tickadoo-mcp/main/connectors/muse/muse.md
Submit at [muse.ai/platform](https://muse.ai/platform) using [`connectors/muse/SUBMISSION.md`](connectors/muse/SUBMISSION.md). No API key. Discovery is this MCP server; checkout stays on tickadoo.com.
## Agent Plugins 1.0
This repository is also a portable Agent Plugins 1.0.0 package. Compatible
clients discover the root [`plugin.json`](plugin.json), the seven workflows in
[`skills/`](skills/), and the credential-free Streamable HTTP configuration in
[`mcp.json`](mcp.json). Current Codex marketplace ingestion uses the parallel
`.codex-plugin/plugin.json` adapter; the portable root manifest remains the
vendor-neutral source of truth.
Run `npm run test:plugin` to validate the package against the vendored official
schemas and its containment, discovery, transport, and secret-safety checks.
The same command verifies the exact npm tarball contents and the provider-neutral
acceptance corpus in [`evals/agent-plugin-scenarios.json`](evals/agent-plugin-scenarios.json).
See [`docs/agent-plugins.md`](docs/agent-plugins.md) for the architecture
decision, compatibility evidence, update/rollback process, and follow-ups.
OpenAI Agents API sessions that call the public MCP (Book path) are documented in
[`docs/openai-agents-api.md`](docs/openai-agents-api.md). Use
`https://mcp.tickadoo.com/mcp` — the bare host 404s.
The package no longer defines tools, formats catalogue data, or calls a local tickadoo backend. It connects to the remote Streamable HTTP MCP server and proxies `tools/list`, `tools/call`, `resources/list`, `resources/read`, and `ping`. The live remote owns the tool list, schemas, results, and errors.
No API key is required.
## Install
Use the hosted remote directly when your MCP client supports Streamable HTTP:
```json
{
"mcpServers": {
"tickadoo": {
"url": "https://mcp.tickadoo.com/mcp"
}
}
}
```
Use the npm package when your MCP client needs a local stdio command:
```json
{
"mcpServers": {
"tickadoo": {
"command": "npx",
"args": ["-y", "@tickadoo/mcp-server"]
}
}
}
```
Use Gemini CLI:
```bash
gemini extensions install https://github.com/tickadoo/tickadoo-mcp
```
Use Claude Code's supported marketplace flow:
```bash
claude plugin marketplace add tickadoo/tickadoo-mcp
claude plugin install tickadoo@tickadoo-agent-plugins
```
## Configuration
Set `TICKADOO_MCP_URL` to point the bridge at another compatible Streamable HTTP MCP endpoint:
```bash
TICKADOO_MCP_URL=http://127.0.0.1:8787/mcp npx -y @tickadoo/mcp-server
```
Set `TICKADOO_LOG_LEVEL=none` to silence bridge status logs on stderr.
## Local Development
```bash
npm install
npm run build
npm test
```
Run the built stdio bridge:
```bash
node dist/index.js
```
Refresh the MCP registry metadata from the live remote:
```bash
npm run sync:server-json
```
Run the optional live integration test:
```bash
LIVE=1 npm test
```
## Live Tools
The current tool list is served by the remote MCP server. Visit [mcp.tickadoo.com](https://mcp.tickadoo.com) or run `npm run sync:server-json` to refresh the registry metadata in this repo.
## Privacy & Data Handling
- **No account or API key required.** The server is read-mostly: it exposes tickadoo's public experiences catalogue (search, recommendations, availability, comparison, itineraries) and returns booking links — it does not collect, store, or require personal data to function.
- **What is sent:** tool arguments (e.g. a city name, query text, or chosen experience id) are forwarded to the tickadoo backend to fulfil the request. The bridge adds no tracking and asks for no credentials.
- **First-party service.** tickadoo is the operator of the catalogue and backend; supplier inventory is presented as tickadoo. Bookings are completed on tickadoo.com.
- **Full policy:** [tickadoo.com/privacy](https://tickadoo.com/privacy). Questions: support@tickadoo.com.
TDQS
Scored across 4 tools
Each tool has a clearly distinct purpose with no overlap: find_nearby_experiences uses geographic location, get_experience_details retrieves details for a specific experience, list_cities enumerates available destinations, and search_experiences searches within a specific city. The descriptions explicitly differentiate their use cases, eliminating any ambiguity.
All tool names follow a consistent verb_noun pattern (e.g., find_nearby_experiences, get_experience_details, list_cities, search_experiences) with clear, descriptive verbs and nouns. There are no deviations in style or convention, making the set highly predictable and readable.
With 4 tools, the set is well-scoped for a ticketing/experience discovery server, covering key functionalities like location-based finding, detailed lookups, city listing, and city-specific searches. Each tool earns its place without redundancy or excessive complexity.
The tool surface covers core discovery and information retrieval workflows effectively, including location-based and city-specific searches, detailed views, and destination listing. A minor gap exists in booking or transaction capabilities, but agents can work around this for recommendation and exploration tasks.