Goose - ScitiX

Goose is Block’s open-source AI coding agent (CLI and desktop). Its OpenAI provider also targets OpenAI-compatible endpoints, so setting a custom host points it at the ScitiX Model Inference API. Default model here is glm-5.2; see the Models page for the full list and pricing.

Configure

Create an API key on the ScitiX API Keys page. Configure the OpenAI provider with a custom host — run goose configure (choose OpenAI, then set the host/base path), or set it directly in ~/.config/goose/config.yaml:

GOOSE_PROVIDER: openai
GOOSE_MODEL: glm-5.2
OPENAI_HOST: https://api.scitix.ai
OPENAI_BASE_PATH: model-api/v1/chat/completions

Provide the API key (stored in the system keyring by goose configure, or via env):

export OPENAI_API_KEY="<Your API Key>"

Field	Description
`GOOSE_PROVIDER`	`openai` — the provider that speaks OpenAI-compatible
`GOOSE_MODEL`	ScitiX model ID, e.g. `glm-5.2`. Goose relies on tool calling, so pick a model with native tool calling.
`OPENAI_HOST`	Host only: `https://api.scitix.ai`
`OPENAI_BASE_PATH`	Full chat-completions path: `model-api/v1/chat/completions`, not just `/v1`
`OPENAI_API_KEY`	Your ScitiX API key

Verify

Run a one-shot prompt — a normal reply confirms the connection (verified on Goose 1.45.0):

goose run --no-session -t "In one short sentence, which model are you?"

The header shows openai glm-5.2 and Goose returns a normal answer, confirming requests reach ScitiX. Goose identifies itself as “goose” regardless of the underlying model. That is its own agent persona, not a routing issue; the session header shows the ScitiX model in use.