Precedence Summary

OpenSteer resolves configuration using this precedence (lowest to highest):

1. Runtime defaults.
2. .opensteer/config.json.
3. Environment variables (including dotenv autoload files, unless disabled).
4. Constructor input passed to new Opensteer(...).

Cloud mode selection is resolved from constructor cloud first, then OPENSTEER_MODE, then default local mode.

6) Configure model and mode (optional)

Opensteer defaults to:

• model: 'gpt-5.1'
• local mode (OPENSTEER_MODE=local)

Set a model:

OPENSTEER_MODEL=gpt-5-mini

Enable cloud mode:

OPENSTEER_MODE=cloud
OPENSTEER_API_KEY=ork_your_key

Additional cloud options:

• OPENSTEER_BASE_URL to override https://api.opensteer.com
• OPENSTEER_AUTH_SCHEME as api-key (default) or bearer
• OPENSTEER_REMOTE_ANNOUNCE as always, off, or tty
• OPENSTEER_CURSOR as true or false

In code, cloud: true or a cloud options object overrides OPENSTEER_MODE. Cloud mode is fail-fast and does not automatically fall back to local mode.

Cursor defaults:

• CLI sessions default to enabled.
• SDK instances default to disabled unless cursor.enabled is set.

7) Dotenv autoload behavior

Opensteer loads .env files from storage.rootDir (default process.cwd()) in this order:

1. .env.<NODE_ENV>.local
2. .env.local (skipped when NODE_ENV=test)
3. .env.<NODE_ENV>
4. .env

Existing process.env values are never overwritten. Set OPENSTEER_DISABLE_DOTENV_AUTOLOAD=true to disable.

The same resolved environment map is used for Opensteer config, CUA provider key resolution, and built-in AI resolve/extract provider setup. Opensteer does not mutate global process.env when loading dotenv files. Env values are captured when an Opensteer instance is created.

Covered Features

• env:OPENSTEER_API_KEY
• env:OPENSTEER_AUTH_SCHEME
• env:OPENSTEER_BASE_URL
• env:OPENSTEER_DISABLE_DOTENV_AUTOLOAD
• env:OPENSTEER_MODE
• env:OPENSTEER_MODEL

Related Docs

• Enable cloud:true
• Environment Variables
• Cloud Errors and Troubleshooting
• Cloud Launch Context