Install & set up

About ten minutes from download to your first approved post.

1. Install

Desktop app: get the installer for your system from Download. It opens in its own window.

Python:

pip install git+https://github.com/jrnoorbaloch/linkedin-agent
linkedin-agent serve          # open http://localhost:8787

Docker: see Self-hosting.

2. Connect a model

Settings → Language model. Pick a provider, paste your key, then click Test connection. Details: Connect any LLM.

3. Import your LinkedIn data

Import → drop LinkedIn's data export ZIP. Your brand profile, profile text, voice samples and post history fill in. See Import your data.

4. Pick a persona pack

Packs → choose the one that fits you. See persona packs.

5. Connect LinkedIn

Create your LinkedIn developer app (step-by-step), add its keys to .env, then Settings → LinkedIn → Connect.

6. Draft, approve, publish

Dashboard → Draft today's post. Review it in the editor, then approve, schedule, or publish now.

Configuration

VariableDefaultPurpose
LA_HOST / LA_PORT127.0.0.1 / 8787Where the portal listens
LA_BASE_URLhttp://localhost:8787Public URL when hosted. Used for OAuth, review links and webhooks
LA_API_TOKENnoneRequires a token for the API. Recommended on servers
LA_DATA_DIR~/.linkedin-agentDatabase, media and encryption key
LA_ACCOUNT1Default workspace for the MCP server
LINKEDIN_CLIENT_ID / _SECRETnoneYour LinkedIn app
LINKEDIN_API_VERSION202608The LinkedIn-Version header
LINKEDIN_ORG_ENABLED0Company Page posting, after approval