Skip to main content
Avala provides official SDKs and client libraries to help you integrate with the platform from your preferred language or framework. Each SDK wraps the Avala REST API with idiomatic methods, built-in authentication, automatic retries, and full type safety.

Client Libraries

Python

Sync and async support. Full type hints. Python 3.9+.

TypeScript

Fully typed. Zero dependencies. Works in Node.js, Deno, and Bun.

CLI

Manage datasets, exports, and storage from your terminal. curl -fsSL https://avala.ai/install.sh | bash.

REST API

Use the API directly from any language with HTTP requests.

OpenAPI

Generate your own client from the OpenAPI specification.

Quick Comparison

FeaturePythonTypeScriptCLIREST API
Installpip install avalanpm install @avala-ai/sdkcurl -fsSL https://avala.ai/install.sh | bashDirect HTTP
AsyncSync + AsyncAsyncN/AAny
Type SafetyFull type hintsFull TypeScript typesN/AN/A
Auto RetriesYesYesVia SDKManual
PaginationBuilt-in iteratorsBuilt-in iterators--limit flagManual
Dependencieshttpx, pydanticNone (native fetch)click, richN/A

Environment Variables

All SDKs respect the following environment variables, so you can configure them once and use them across your entire stack.
VariableDescriptionDefault
AVALA_API_KEYYour Avala API key for authentication. Used automatically if no key is passed to the client constructor.None (required)
AVALA_BASE_URLOverride the default API base URL. Useful for self-hosted deployments or development environments.https://api.avala.ai/api/v1

Authentication

All requests to the Avala API must include your API key in the X-Avala-Api-Key header. You can find your API key in the Avala dashboard under Settings > Security.
export AVALA_API_KEY="your-api-key"
Once the environment variable is set, all SDKs will pick it up automatically without any additional configuration.

MCP Integration

Use Avala with AI assistants like Claude, Cursor, and VS Code through the Model Context Protocol. The @avala-ai/mcp-server package exposes your datasets, projects, and exports as tools that AI assistants can call directly.