Skip to main content

FAQ

An MCP server factory and skill studio: paste an API spec, get a typed MCP server deployed to a global edge runtime, and author reusable agent skills on top of it.
Yes. The generated TypeScript server is portable and the deployable output remains yours. Tool descriptions stay editable, and secrets never land in generated source.
OpenAPI/Swagger, GraphQL (via introspection), GitHub repositories, and hosted spec URLs. You curate which operations become tools.
On a global edge runtime (Cloudflare Workers) with sub-5ms cold starts, distributed across 300+ locations. Fly.io and self-hosted targets are planned.
API (control-plane) auth uses long-lived personal API keys for managing projects, servers, and deployments. Runtime auth uses short-lived tokens to call a protected deployed server. See Runtime access.
A server exposes tools generated from an API. A skill is a reusable, versioned agent behavior that composes those tools into a workflow with triggers and guardrails. See Skill Studio.
No — the dashboard covers the full flow. The CLI is recommended for repeatable workflows and CI/CD. mcp, mcpsh, and mcpcloud are the same binary, and every command supports --json.
Every server ships Streamable HTTP (default), SSE (legacy clients), and stdio (local dev). Most remote clients use Streamable HTTP. See Transports & auth.
Credits meter platform work like AI enrichment. Each plan includes a monthly allotment that resets per billing period; paid tiers can buy prepaid packs. See Credits & plans.
Yes — publish servers and skills to the marketplace. Monetized artifacts return purchase_required (402) until purchased, and creators keep 80% of revenue.
Read What is MCP? and What MCPCloud provides, then use the API Reference or CLI Reference to automate.