DeepSeek Harness

2. Configure

DeepSeek Harness keeps model providers in ~/.dsh/settings.yaml (or $DSH_HOME/settings.yaml). Set the API Key environment variable, then add the GetRun provider:

Windows

$env:GETRUN_API_KEY = "sk-xxxxx-xxx"
$dshHome = if ($env:DSH_HOME) { $env:DSH_HOME } else { Join-Path $HOME ".dsh" }
New-Item -ItemType Directory -Force -Path $dshHome | Out-Null
notepad (Join-Path $dshHome "settings.yaml")

macOS

export GETRUN_API_KEY="sk-xxxxx-xxx"
DSH_HOME="${DSH_HOME:-$HOME/.dsh}"
mkdir -p "$DSH_HOME"
open -e "$DSH_HOME/settings.yaml"

Linux

export GETRUN_API_KEY="sk-xxxxx-xxx"
DSH_HOME="${DSH_HOME:-$HOME/.dsh}"
mkdir -p "$DSH_HOME"
${EDITOR:-vi} "$DSH_HOME/settings.yaml"

Write into settings.yaml:

settings.yaml

llm-pi-ai:
  providers:
    getrun:
      displayName: GetRun
      apiKeyEnv: GETRUN_API_KEY
      api: openai-completions
      baseURL: https://api.get-apikey.com/v1
      models:
        - id: YOUR_MODEL
          name: YOUR_MODEL

Replace sk-xxxxx-xxx and YOUR_MODEL with your actual values. The change takes effect on the next request; no restart is required. You can also save the API Key directly on the Models page in dsh web instead of setting the environment variable.

3. Use

Run dsh web, fill in the API Key on the model configuration page first, then select YOUR_MODEL.

4. Confirm a reply

Start a new conversation and send Reply only with OK. The setup is complete when DeepSeek Harness replies OK.