Skip to main content

Prerequisites

Install

The @cognition-ai packages are in beta. Every release is a 0.0.1-beta.N prerelease on the beta dist-tag (currently 0.0.1-beta.7), so the @beta suffix is required: a bare npm install @cognition-ai/sdk fails on purpose. Pin the exact version in your lockfile, and keep every @cognition-ai package you use (sdk, ai-sdk, cloudflare-agents, harness-devin) on the same version; they release together. Each published version is stable, but the API can change between beta bumps; see the changelog for what moved.
Using a coding agent? Install the devin-sdk skill and it will know how to use this SDK:

Run a session

createDevin() connects to the cloud. Pass cwd to run the bundled Devin CLI on your machine instead. Both need an API key, from the apiKey option or DEVIN_API_KEY; nothing else in the environment changes which one you get.
Clients are async-disposable, so await using closes the connection (and the CLI process) at scope exit. Call .close() yourself if you don’t use it.

Next steps

Stream events

Print the reply as it arrives and react to each tool call.

Structured output

Get a typed, validated value back instead of prose.

How it works

Run the same code in the Devin CLI on your machine with createDevin({cwd}).

Using devin.api

Read the session’s PRs and structured output, or manage playbooks and secrets.

Examples

Five runnable programs, including this quickstart with a --local flag.

SDK reference

Every exported class, function, and type, generated from the source.