Use Palepale in DeepSeek Harness

Install DeepSeek Harness Desktop. Windows x64 · Mac with Apple Silicon · Mac with Intel

Install Palepale CLI v0.1.71 or later. If you already have it, run palepale update. Palepale CLI setup

Cockpit sync requires one confirmed Palepale wallet purchase of at least $5. In Cockpit, switch to the Codex account you want to use so it is marked Current, then run:

palepale deepseek

Approve the Palepale sign-in if prompted. The launcher checks that the local Codex login matches Cockpit's current account and imports it into a private Harness profile. API Service and a separate API key are not required. Choose your project in the desktop app; GPT 6.1 Sol is the default. Use the model picker or run palepale deepseek MODEL_ID to choose another model.

Cockpit requests use that account's provider access without spending your Palepale wallet balance. Run palepale deepseek --palepale to use Palepale API billing instead, with DeepSeek V4.1 Flash 1017 selected when available. Check Palepale usage.

After switching accounts in Cockpit, quit DeepSeek Harness Desktop and run palepale deepseek again. A changed account or source login opens a separate profile; previous conversations stay in their original profile. Reopening the same login preserves tokens already refreshed by Harness.

Credentials stay on your computer and go directly to the provider. Harness refreshes its own login; it does not write refreshed tokens back to Cockpit. If sign-in is needed again, reconnect the account in Cockpit and relaunch.

For the web app, install Node.js 24.9 or later and run npm install -g @deepseek-ai/dsh@0.2.2-rc.1. The command opens the web app by default on Linux. Use palepale deepseek --web to open it in a browser on Windows or macOS.

Need a hand?

Open Palepale Discord for account support. Compare model rates or read about wallet billing.