Using SillyTavern with SoAI

SillyTavern can chat through SoAI's OpenAI-compatible API. SillyTavern does not provide a native Streamable HTTP MCP client, so this integration covers model inference.

Connect the model API and understand MCP limits

In the SoAI WebUI, select the Settings button at the right of the header, choose Settings, and open OpenAI. This tab appears only to an Administrator with API key management (OPENAI_API_ADMIN). Select Create key and copy the key when it is shown. Open Models and copy the exact ID of a model that supports Chat Completions; this requires Model read access (MODEL_READ).

Connect through the SillyTavern interface

  1. Open API Connections.
  2. Set API type to Chat Completion.
  3. Set Chat Completion Source to Custom (OpenAI-compatible).
  4. Enter https://SOAI_HOST:5090/v1 as the custom endpoint and paste the SoAI API key. Do not append /chat/completions; SillyTavern adds the endpoint path.
  5. Refresh the model list and select an exact model ID. If discovery is unavailable, enter the ID manually.
  6. Select Test Message, then save the connection as a Connection Profile if it should be reused.

Generic compatibility does not imply every SillyTavern sampler or extension field is accepted. Use controls supported by the selected model and inspect server logs if a request is rejected.

MCP boundary

Do not add /mcp as a chat endpoint. It is not an OpenAI-compatible inference route. Use a documented SillyTavern extension or a separate gateway when tools are required.

Related documentation