agent-bom discover gcp

This Skill addresses the challenge of uniformly inventorying AI Agent and MCP-related assets in GCP environments. In practice, teams often need to manually log into the GCP Console, query resources service by service (Vertex AI, Cloud Run, Cloud Functions, GKE), and then manually compile a list—a tedious process prone to omissions. By calling GCP's read-only APIs, this Skill automatically discovers all AI-related assets and outputs a standardized JSON inventory compliant with the agent-bom specification, significantly improving asset inventory efficiency.

Usage is straightforward: simply ensure Python 3.11+ and the agent-bom library are installed locally, and that GCP Application Default Credentials (ADC) or Workload Identity are properly configured. Then, instruct your AI assistant with "Inventory the AI Agent and MCP assets in the current GCP project." It will automatically execute the discovery process and write the results to your specified JSON file. The entire process requires no manual input of GCP credentials, as they remain securely in your local environment.

It is ideal for teams and individuals who need to regularly audit GCP AI assets, perform security and compliance checks, or plan migrations. Users already leveraging the agent-bom ecosystem for asset inventory management can directly import the generated JSON into subsequent scanning or analysis workflows. Operations engineers, agent developers, and cloud architects can all benefit.

It is recommended to configure ADC via gcloud auth application-default login before first use and ensure the current project has sufficient read-only permissions. Note that this Skill only performs discovery operations and will not modify any GCP resources. Sensitive information (such as service account keys) is automatically redacted in the generated inventory. For projects with a large number of assets, it is advised to execute the Skill per project to avoid API rate limiting.