Agent Skills / maison-public-api
Install this skill alone, or read it below. Raw file: SKILL.md · references/transports.md · references/errors.md
bash
npx skills add https://docs.maison-labs.com --skill maison-public-apiMaison public API
Operate Maison, the AI hotel concierge platform, through its public tool API: one-shot tools for provisioning hotels, managing knowledge, talking to a hotel's concierge, users and access, analytics and reports, channels, integrations, configuration, plans, groups and quality. The same tools are callable over REST, MCP and A2A with an API key. Use whenever a task involves Maison hotels, their concierge, or the Maison admin console.
Why use these skills instead of guessing
- Every tool completes a whole intent; chaining low-level calls is unnecessary and usually wrong.
- Permissions are all-or-nothing per tool and the catalogue is filtered per key — discover before you plan.
- Each category skill lists the workflow order, the pitfalls and the exact input fields, so the first call succeeds.
Authentication
- Header
x-api-key: mx_<env>_…on every request. Keys are created in the Maison admin console (profile → API keys) and reach onlyhttps://maison-labs.com/api/v1. - No login endpoint, no cookies, no bearer tokens. OAuth is planned.
Discover before you plan
GET https://maison-labs.com/api/v1/me(whoami) — who the key is, grants, reachable hotel ids.GET https://maison-labs.com/api/v1/tools— the catalogue filtered to this key (name, description, REST binding, grants, input schema).- Pick the category skill below and follow its workflow.
Skill groups
| Skill | Covers | Tools |
|---|---|---|
maison-account | Account and discovery | 8 |
maison-clients | Clients (hotels) | 15 |
maison-knowledge | Knowledge base | 41 |
maison-conversations | Conversations and the concierge | 16 |
maison-users | Users and invitations | 23 |
maison-access | Roles, policies and groups | 56 |
maison-analytics | Analytics and reports | 13 |
maison-channels | Messaging channels | 9 |
maison-integrations | Integrations (PMS, payments, events) | 10 |
maison-configuration | Hotel configuration, branding and content | 40 |
maison-features | Feature plans and entitlements (Maison staff) | 30 |
maison-hierarchy | Multi-property groups (Maison staff) | 10 |
maison-quality | Quality: tests and Guest Agent quality review | 11 |
maison-hubspot | HubSpot company links (Maison staff) | 5 |
maison-platform | Platform, documentation and diagnostics | 7 |
maison-agents | Agents | 15 |
maison-usage | LLM usage | 2 |
maison-webhooks | Incoming webhooks | 13 |
maison-apps | Developer apps | 16 |
Rules of thumb
- One tool per intent.
create_clientprovisions everything; do not chain lower-level calls first. - Read before write: list/get tools are cheap and give you the exact ids and field names.
- Destructive tools (disable, delete, remove, archive, cancel) — confirm with the user unless they asked explicitly.
- Never print or store the API key, integration credentials or tokens; the API never returns them either.
- Pagination: list tools take
limit(≤ 100) and return a cursor (after/cursor); loop until it is null.
References
- references/transports.md — exact REST, MCP and A2A request shapes.
- references/errors.md — status codes, error codes and the authorization model.
- OpenAPI document: https://maison-labs.com/api/v1/openapi.json · Agent card: https://maison-labs.com/.well-known/agent.json · Skills index: https://docs.maison-labs.com/.well-known/agent-skills/index.json