Skip to main content
Gcore Developer Tools manage services programmatically using the REST API, SDKs, Terraform, or an MCP server. Choose the approach that fits the workflow, from direct API requests and application integrations to infrastructure as code and AI-assisted operations. All four approaches use a Gcore API token for authentication.

​
Direct requests

HTTP fits a single command, a shell script, or a language without a published client. One token reaches Cloud, CDN, DNS, Storage, FastEdge, and the other products. The REST API uses one request shape, one error body, and one page size on every product, and a create or delete call returns a task to poll when the work is not finished in the response. The first call confirms the token and returns the project and region IDs that later calls need. The API reference lists each operation with its parameters and response fields.

​
Application code

Call generated methods from Python and Go instead of assembling URLs, because the method and parameter names stay aligned with the OpenAPI specification. The SDK overview names the packages, the language versions they require, and the cases where a raw request is still the better fit. JavaScript services use fetch or axios against the same endpoints, because no JavaScript client is published.

​
Infrastructure as code

Terraform stores Cloud, CDN, DNS, and WAAP resources as configuration that can be reviewed, reapplied, and rolled back. Provider v2 tracks the current specification and is the version for a new project, while an existing project stays on v0 until its configuration is migrated. The provider overview shows the difference in authentication and resource syntax. Getting started installs the CLI and applies a first file. A resource that already exists in the account enters that file through import without being created again.

​
AI assistants

An assistant that speaks the Model Context Protocol can provision a resource or read its state from a plain-language request, then chain the lookups that request needs. The MCP server exposes those operations as tools for Claude Code, Cursor, and MCP Inspector. Server setup connects the server to a client with an API token. Clients that load every tool schema at once stay within their context limit through tool filtering.