DeepSeek Harness

Connect DeepSeek Harness (dsh) to GetVibe API: Custom model API, the OpenAI Chat Completions protocol, base URL getvibeapi.com/v1.

DeepSeek Harness (dsh) connects as a separate “Custom model API” provider over the OpenAI Chat Completions protocol. The built-in DeepSeek card only talks to api.deepseek.com — do not paste the key there.

Web UI setup

Open Settings → Models → Add model provider and switch the card to Custom model API:

Parameter	Type	Description
`Provider ID`	field	vibeapi — lowercase; it cannot be changed later
`Base URL`	field	https://getvibeapi.com/v1 — exactly, with no /chat/completions at the end
`API protocol`	select	OpenAI Chat Completions
`API key`	field	Your sk-pv-… key from the dashboard
`Model catalog`	list	Fetch available models — or type deepseek-v4.1-flash by hand

The Base URL ends in /v1: the harness appends /chat/completions itself. An address that already contains that path turns into /v1/chat/completions/chat/completions.

Through the config file

The same in ~/.dsh/profiles/web/cordis.patch.yml. Put the key as the line GETVIBE_API_KEY=sk-pv-… in ~/.dsh/.env. If you already have an llm-pi-ai entry, add the provider inside it — the entry replaces the whole block.

- id: llm-pi-ai
  config:
    providers:
      vibeapi:
        apiKeyEnv: GETVIBE_API_KEY
        api: openai-completions
        baseURL: https://getvibeapi.com/v1
        models:
          - id: deepseek-v4.1-flash
          - id: deepseek-v4-pro-0813

Verify

Pick the model in the harness model picker and send “Reply: ready”. Other ids are in the catalog.

← PreviousHermes Agent Next →Open WebUI