agent-bom discover gcp
Inventorying AI Agent and MCP Assets on GCP
Install & Use
Copy this prompt and send it to your AI assistant (Claude / Cursor / TRAE / Codex / WorkBuddy etc.) to auto-install:
Help me install this AI Skill: agent-bom discover gcp. It is used for: Inventorying AI Agent and MCP Assets on GCP Full Skill content: https://321skill.com/skills/agent-bom-discover-gcp-x/raw/index.md Read that page and install it.
The prompt includes a link to the full Skill content. You can also view the full content.
This Skill addresses the problem of quickly discovering AI Agent and MCP-related assets within a GCP environment and generating a standardized inventory. Operations and development personnel often need to manually query multiple GCP services (such as Vertex AI, Cloud Run, Cloud Functions, GKE) to obtain asset information. This process is cumbersome, prone to omissions, and lacks a unified format.
Usage is straightforward. You simply need to install the agent-bom tool locally and configure read-only GCP credentials (ADC, Workload Identity, or a temporary service account). Afterwards, you can use natural language instructions to automatically discover AI-managed resources in your projects, output a standardized agent-bom inventory JSON file, and support local scanning of the inventory.
It is ideal for teams or individuals managing AI infrastructure on GCP, especially for operations engineers and agent developers who have already deployed multiple AI Agents, microservices, or Kubernetes clusters. It helps them quickly gain a comprehensive overview of their assets, providing foundational data for subsequent audits, cost optimization, or security scanning.
It is recommended to run this Skill after each major change or periodically to update the asset inventory. Note that it only performs discovery and scanning operations; it does not modify any GCP resources. Credentials always remain within the operator's environment and are not exposed to third parties.
Key Features
Unlike manually querying services one by one using `gcloud` commands, it automatically discovers AI Agent and MCP-related assets and outputs a standardized `agent-bom` JSON format. It does not require long-term credentials, instead leveraging the operator's existing temporary credentials (ADC), making it more secure and easier to integrate into CI/CD pipelines.
Limitations
Requires a Python 3.11+ environment and must be run on a machine already configured with GCP read-only credentials (ADC, Workload Identity, or a temporary service account). It does not support environments without the GCP SDK.
FAQ
Does agent-bom require long-term GCP credentials?
No. It uses the operator's existing Application Default Credentials, Workload Identity, or short-term service account credentials. Credentials never leave the operator's environment.
Does it support multiple GCP projects?
Yes. By specifying the `--project` parameter or setting the `GOOGLE_CLOUD_PROJECT` environment variable, you can run it for a single project. You can also run it multiple times to cover multiple projects.
Installation guide for AI assistants
If your AI coding assistant (Claude Code, Cursor, TRAE etc.) can see this page, send it this message to auto-install:
Visit https://321skill.com/skills/agent-bom-discover-gcp-x/raw/index.md to read the original Skill definition (Markdown format) for agent-bom discover gcp, and install it according to the instructions.
Raw Markdown URL for AI: /skills/agent-bom-discover-gcp-x/raw/index.md