Agent Skills / maison-platform

Install this skill alone, or read it below. Raw file: SKILL.md · references/tools.md

bash

npx skills add https://docs.maison-labs.com --skill maison-platform

Maison — Platform, documentation and diagnostics

Orient and verify: the documentation index with machine-readable discovery URLs, the exact widget embed code for a hotel, a website check (framework, widget present, client id), the AI site diagnostic (start, read findings, ask follow-ups) and a live platform health summary. Use when installing the widget on a hotel's site, when a site seems not to load the concierge, or to find the right documentation page.

Part of the Maison public API. Read the maison-public-api skill first for authentication, discovery and transport details (https://docs.maison-labs.com/skills/maison-public-api/SKILL.md).

When to use

Install the widget — get_widget_embed_code, then check_website to verify.
A hotel site misbehaves — start_site_diagnostic → get_site_diagnostic → chat_site_diagnostic.
Find documentation — get_documentation.

Before you start

Send x-api-key: <key> on every call; the key only reaches /api/v1.
GET /api/v1/tools (or MCP tools/list) returns the tools THIS key may run — if a tool below is missing there, you lack a grant; do not retry it.
Client-scoped tools take clientId (a hotel id). whoami lists the ids you can act on.

Tools

Tool	What it does	REST	Needs
`chat_site_diagnostic`	Send a follow-up question about a finished or running diagnostic session and get the agent's answer.	`POST /api/v1/diagnostic/{sessionId}/chat`	a valid API key
`check_website`	Fetch a public website and report its framework, whether the Maison widget loader is embedded and which client id it carries.	`POST /api/v1/tools/sitecheck`	a valid API key; read-only
`get_documentation`	Where the public documentation lives: the docs site topics (quick start, authentication, API reference, MCP, A2A, widget SDK, installation guides, integrations, diagnostic), plus the machine-readable discovery URLs (OpenAPI document, agent card, filtered tool catalogue).	`GET /api/v1/docs`	a valid API key; read-only
`get_platform_status`	A live health summary from the API's point of view: environment, database reachability, LLM service reachability and the public status page URL.	`GET /api/v1/status`	a valid API key; read-only
`get_site_diagnostic`	The full history of a diagnostic session — run events, progress, findings and chat turns.	`GET /api/v1/diagnostic/{sessionId}`	a valid API key; read-only
`get_widget_embed_code`	The exact HTML a hotel pastes into its website to show the Maison chat widget: the hardened loader <script> tag for this environment's CDN and the `initMcbs` call with the hotel's client id.	`GET /api/v1/client/{clientId}/embed`	clients READ (client); read-only
`start_site_diagnostic`	Start the AI website diagnostic for a URL: it crawls the site, checks the widget, content and structure, and streams findings.	`POST /api/v1/diagnostic`	a valid API key

Full input fields for every tool: references/tools.md.

Recommended workflows

Install and verify the widget

get_widget_embed_code for the hotel; give the operator the html block.
After they publish, check_website with the site URL: hasAgentScript and the detected client id confirm the install.
If something is off, start_site_diagnostic and read the findings.

Tips

get_platform_status is a quick way to tell an API outage from a hotel-specific problem.

Pitfalls

Diagnostics run against public URLs only; private or password-protected sites cannot be checked.

Calling conventions

Same tool, three transports — pick whichever your runtime already speaks. Always send x-api-key.

bash

# REST
curl -X POST "https://maison-labs.com/api/v1/diagnostic/<sessionId>/chat" -H "x-api-key: $MAISON_API_KEY" -H "content-type: application/json" -d '{ "message": … }'

json

// MCP (https://maison-labs.com/mcp, Streamable HTTP): tools/call
{ "jsonrpc": "2.0", "id": 1, "method": "tools/call", "params": { "name": "chat_site_diagnostic", "arguments": { "message": … } } }

json

// A2A (https://maison-labs.com/a2a): tasks/send with a data part
{ "jsonrpc": "2.0", "id": 1, "method": "tasks/send", "params": { "message": { "role": "user", "parts": [{ "type": "data", "data": { "tool": "chat_site_diagnostic", "input": { "message": … } } }] } } }

Errors are the same everywhere: 401 no/invalid key; 403 forbidden with details.missing (the grants you lack — fall back to the composite's parts or stop); 403 not_entitled (the hotel's plan lacks the feature); 404, 409, 413 too_large, 503. Inputs that fail validation return 400 invalid_input with the Zod issues — fix and retry once.