Skip to main content
Every step below shows the same action on each surface - pick one tab and stay in it end to end:
  • Python / TypeScript - the SDKs, for when Engini is called from your code
  • CLI - the engini terminal tool (npm), for shells, scripts, and CI
  • cURL - the raw REST API, for any other language
All four hit the same API with the same key, so you can switch lanes later without redoing anything.

1. Get an API key

Create a Developer API key on your Engini account’s keys page. Keys are prefixed eng_ and are sent in the x-api-key header - the key is bound to your account server-side, so it’s the only credential you need.
Never commit API keys. Export them as environment variables: ENGINI_API_KEY.

2. Install

3. Verify your key

4. Discover applications and tools

5. Connect an application

Tools execute through a connection - your authenticated link to the application (an OAuth grant, an API key you paste, DB credentials). You create it once; every surface can use it afterwards.
This is the one step where we recommend the CLI whatever lane you’re in: engini connect is a guided flow that handles credentials, OAuth, and object selection in one go. A connection created here is immediately usable from the SDKs and the REST API - it lives in your account, not in the CLI.

6. Execute a tool

A 200 response does not mean the tool succeeded - always check isSuccess. Tool-runtime failures (e.g. the downstream API rejected the call) return 200 with isSuccess: false and an errorMessage; every other failure uses the standard error envelope.

Next steps

Connections & OAuth

Create connections programmatically, run the OAuth flow, select objects.

Give tools to an LLM

Toolsets + OpenAI/Anthropic providers - the agent loop.

CLI machine contract

Exit codes, JSON output, --schema, result handles.

For AI Agents

How agents should discover and operate Engini.