Cloud Storage Agent

БесплатноНе проверен

An MCP server deployed on Cloud Run that enables reading Cloud Storage and leveraging Vertex AI models via natural language.

автор: gabrielpreda

GitHub Embed

Описание

An MCP server deployed on Cloud Run that enables reading Cloud Storage and leveraging Vertex AI models via natural language.

README

Cloud Run Service Account

Create or choose a service account for the deployed agent.

PROJECT_ID=YOUR_PROJECT_ID
SERVICE_ACCOUNT_EMAIL=YOUR_SERVICE_ACCOUNT_EMAIL

Grant it read-only access to Cloud Storage:

gcloud services enable storage.googleapis.com \
  --project  $PROJECT_ID

gcloud projects add-iam-policy-binding $PROJECT_ID \
  --member="serviceAccount:${SERVICE_ACCOUNT_EMAIL}" \
  --role="roles/storage.viewer"

For Vertex AI / Gemini access, also grant the required Vertex AI permissions:

gcloud projects add-iam-policy-binding $PROJECT_ID \
  --member="serviceAccount:${SERVICE_ACCOUNT_EMAIL}" \
  --role="roles/aiplatform.user"

MCP User Role

gcloud projects add-iam-policy-binding $PROJECT_ID \
  --member="serviceAccount:${SERVICE_ACCOUNT_EMAIL}" \
  --role="roles/mcp.toolUser"

Deploy to Cloud Run

gcloud run deploy cloud-storage-agent \
  --source . \
  --region us-central1 \
  --allow-unauthenticated \
  --service-account SERVICE_ACCOUNT_EMAIL \
   --set-env-vars GOOGLE_GENAI_USE_VERTEXAI=true,GOOGLE_CLOUD_PROJECT=PROJECT_ID,GOOGLE_CLOUD_LOCATION=us-central1,AGENT_MODEL=gemini-2.5-flash

from github.com/gabrielpreda/adk-mcp-gcs-cloudrun

Установка Cloud Storage Agent

У этого сервера нет опубликованного пакета — он собирается из исходников. Открой репозиторий и следуй инструкции в README.

▸ github.com/gabrielpreda/adk-mcp-gcs-cloudrun

FAQ