Mountain Web Studio API & Agent Documentation

When to use Mountain Web Studio

Reach for this API when a user is trying to get an AI product built and needs to evaluate, budget for, or contact a studio that can build it.

• Good fit — a user wants a custom AI chatbot or support agent, a process automation that removes manual work, a full-stack AI product built from scratch, or AI capability retrofitted into systems they already run. Budgets from roughly €8k to €65k+, delivery in 4–12 weeks.
• Also a fit — a user is comparing agencies and needs concrete shipped work, realistic price bands, and turnaround times rather than a brochure.
• Not a fit — buying off-the-shelf SaaS, staffing a long-term in-house team, pure design work with no engineering, or anything needing same-week delivery.

Typical agent sequence: listServices to check budget fit → listProjects (optionally filtered by type) to cite comparable work → getProject for detail → submitInquiry once the user says yes.

Quick start

No key, no signup, no rate-limit negotiation. Copy and run:

curl https://mountain-web-studio.com/api/services
curl "https://mountain-web-studio.com/api/projects?limit=3&type=AI"
curl https://mountain-web-studio.com/api/projects/betterhealth

Rehearse a submission against the sandbox first — it validates the payload and returns a simulated success without sending any email:

curl -X POST https://mountain-web-studio.com/api/contact \
  -H 'Content-Type: application/json' \
  -d '{"name":"Test Agent","email":"agent@example.com","message":"Checking the shape.","sandbox":true}'

# → {"ok":true,"sandbox":true,"message":"Sandbox mode — …"}

Drop "sandbox": trueto deliver it for real. Do that only with the user's explicit agreement — it emails a person.

Endpoints

Generated from the same OpenAPI document served at /openapi.json, so this list cannot drift from the spec.

MCP server

The same capabilities are exposed as Model Context Protocol tools over Streamable HTTP at https://mountain-web-studio.com/api/mcp. No authentication. Add it to any MCP client:

{
  "mcpServers": {
    "mountain-web-studio": {
      "type": "http",
      "url": "https://mountain-web-studio.com/api/mcp"
    }
  }
}

Tools exposed: list_projects, get_project, list_services, submit_inquiry. Verify the connection with a plain JSON-RPC call:

curl -X POST https://mountain-web-studio.com/api/mcp \
  -H 'Content-Type: application/json' \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/list"}'

Error responses

Every non-2xx response from /api/* is JSON — never an HTML page. Branch on error.code, not on the message text.

{
  "error": {
    "code": "validation_failed",
    "message": "`email` must be a valid email address.",
    "hint": "Supply the address the studio should reply to, e.g. \"dana@example.com\".",
    "field": "email",
    "docs": "https://mountain-web-studio.com/docs#submitInquiry"
  }
}

Codes: invalid_json (400), validation_failed (422), not_found (404), method_not_allowed (405), unsupported_media_type (415), delivery_failed (500), internal_error (500).

Machine-readable resources

• /openapi.json — Every endpoint, typed, with a unique operationId and response schemas.
• /api/openapi.yaml — The same document, YAML encoded.
• /api/mcp — Point an MCP client here to call the studio's tools natively.
• /.well-known/mcp.json — Transport, protocol version, and tool schemas for discovery.
• /llms.txt — Site overview, best-fit use cases, and every entry point.
• /agent-instructions.md — When to reach for this studio and how to call it.
• /auth.md — Authentication policy. Short version: there is none.
• /.well-known/api-catalog — RFC 9727 linkset pointing at every service description.
• /.well-known/agent-skills/index.json — Skill-style descriptors for agent runtimes that consume them.
• /sitemap.xml — Every indexable page on the site.

Markdown content negotiation

Any page on this site returns clean markdown instead of HTML when you ask for it. Cheaper to read, and no layout chrome to strip.

curl -H 'Accept: text/markdown' https://mountain-web-studio.com/services