Skip to content

vertex.md

Latest commit

 

History

History
69 lines (50 loc) · 2.08 KB

vertex.md

File metadata and controls

69 lines (50 loc) · 2.08 KB
Raw

Integration Guide: Google Vertex AI (Gemini)

Vertex AI Gemini models use the same tool schema as the Gemini API. Use SkillLoader.to_gemini_tool() — no separate Skillware adapter. Auth and client init differ from the consumer API key path.

Install

pip install "skillware[gemini]"
pip install "skillware[<category>_<skill>]"

Client pattern

Use Application Default Credentials (ADC) on GCE, GKE, or Cloud Run, or set GOOGLE_APPLICATION_CREDENTIALS for a service account key file.

import os

import google.genai as genai
from google.genai import types

from skillware.core.env import load_env_file
from skillware.core.loader import SkillLoader

load_env_file()

bundle = SkillLoader.load_skill("finance/wallet_screening")
skill = bundle["class"](
    config={"ETHERSCAN_API_KEY": os.environ.get("ETHERSCAN_API_KEY")}
)
tool = SkillLoader.to_gemini_tool(bundle)

client = genai.Client(
    vertexai=True,
    project=os.environ["GOOGLE_CLOUD_PROJECT"],
    location=os.environ.get("GOOGLE_CLOUD_LOCATION", "us-central1"),
)

response = client.models.generate_content(
    model=os.environ["VERTEX_GEMINI_MODEL"],
    contents="Screen wallet 0xd8dA6BF26964aF9D7eEd9e03E53415D37aA96045.",
    config=types.GenerateContentConfig(
        tools=[tool],
        system_instruction=bundle["instructions"],
    ),
)

Handle function_call parts and skill.execute() as in gemini.md.

Environment

Variable Purpose
GOOGLE_CLOUD_PROJECT GCP project ID
GOOGLE_CLOUD_LOCATION Region (for example us-central1)
VERTEX_GEMINI_MODEL Vertex model resource ID
Skill env_vars Separate — see catalog page

Hosting

Run Skillware on Compute Engine, GKE, or Cloud Run — pip install skillware, attach a service account with Vertex access, use ADC. See enterprise_cloud.md.

Related