Use uncensored models in DeepSeek Harness

Add Silk Compute as a provider in ~/.dsh/settings.yaml, set SILK_API_KEY in your shell, and pick a model under Settings, then Models.

Setup

~/.dsh/settings.yaml

llm-pi-ai:
  providers:
    silkcompute:
      displayName: Silk Compute
      apiKeyEnv: SILK_API_KEY
      api: openai-completions
      baseURL: https://api.silkcompute.com/v1
      models:
        - id: kimi-k3.1-uncensored
          contextWindow: 1048576
          maxTokens: 943718
        - id: deepseek-v4.1-flash-1017-uncensored
          contextWindow: 1048576
          maxTokens: 393216
        - id: glm-5.3-flash-uncensored
          contextWindow: 1048576
          maxTokens: 131072

apiKeyEnv takes the name of the environment variable, not the key itself. To use the app instead, open Settings, then Models, then Add a custom provider, and enter the same base URL, the openai-completions protocol, and SILK_API_KEY.

DeepSeek Harness is in developer preview, so field names can change between versions. If yours differ, check the README that ships with it.