remote/vertexai

Uni-Xervo support

• Provider ID: remote/vertexai
• Feature flag: provider-vertexai
• Capabilities: embed, generate

Authentication and project

Defaults used by Uni-Xervo:

• token env var: VERTEX_AI_TOKEN
• project fallback env var: VERTEX_AI_PROJECT

Uni-Xervo provider options

• api_token_env (string)
• project_id (string)
• location (string)
• publisher (string)
• embedding_dimensions (integer > 0, embed task only)

Authoritative Uni-Xervo option schema:

• https://github.com/rustic-ai/uni-xervo/blob/main/schemas/provider-options/vertexai.schema.json

Authoritative model and config docs

• Model catalog: https://cloud.google.com/vertex-ai/generative-ai/docs/models
• Generation model reference/config: https://cloud.google.com/vertex-ai/generative-ai/docs/model-reference/inference
• Embeddings API/config: https://cloud.google.com/vertex-ai/generative-ai/docs/model-reference/text-embeddings-api

Uni-Xervo generation options exposed

• max_tokens
• temperature
• top_p

Example catalog entry

{
  "alias": "embed/vertex",
  "task": "embed",
  "provider_id": "remote/vertexai",
  "model_id": "text-embedding-005",
  "options": {
    "project_id": "my-gcp-project",
    "location": "us-central1",
    "api_token_env": "VERTEX_AI_TOKEN"
  }
}