LabscoMCP early access ↗
spolspolMCP SERVER

Terragrunt GCP MCP Tool

Plan, apply and roll back Terragrunt-managed GCP infrastructure — with dependency graphs and cost analysis alongside.

by spolspol ↗Cloud Resources & Infrastructure as CodeCommunity

LABSCO SUMMARY

Visualise, validate, then apply — in that order.

What lifts this above a Terragrunt wrapper is the read side: a dependency graph you can render as Mermaid, a drift check, and cost forecasting from real billing data all sit next to the apply. That gives an assistant a way to understand the estate before changing it, which matters here because the same credentials that draw the tree can also delete from it.

What it is

A Terragrunt operations server for Google Cloud. It reads the infrastructure, draws its dependencies, plans and applies changes, checks for drift, and reports on what it all costs — including support for Terragrunt's experimental stacks, which run units in parallel with dependency ordering.

What you get

  • draw_resource_tree, get_dependency_graph in DOT, Mermaid or JSON, and visualize_infrastructure combining both
  • plan_resource_deployment then apply_resource_deployment with safety checks, check_deployment_status while it runs, rollback_deployment when it goes wrong
  • validate_resource_config checks a configuration with its dependencies before anything is applied
  • check_drift detects configuration drift; get_infrastructure_status and analyze_dependencies cover the wider read
  • Stacks: list_stacks, get_stack_details with unit execution order, execute_stack_command for parallel runs, get_stack_outputs for aggregated outputs
  • Cost tools against the Google Cloud Billing API — get_cost_analysis with forecasting, get_cost_alerts on budget thresholds, get_cost_optimization_score, get_cost_status
  • create_resource, update_resource and delete_resource, the last with dependency checking
  • send_slack_notification, create_deployment_summary and get_audit_log for the team side
  • get_autodevops_system_prompt returns a ready system prompt in compact, extended or CLI variants

Requirements

Python, installed with pip install -e . against the repository. Copy config/config.example.yaml to config/config.yaml and fill it in — the server takes that path as its argument. GCP access comes from GOOGLE_APPLICATION_CREDENTIALS pointing at a service account key, which is what the apply and delete tools act with, so scope that account deliberately. Everything the same CLI does is also available as python3 -m terragrunt_gcp_mcp.cli. Version 0.3.3.

Setup effort

Build from source — git clone https://github.com/spolspol/terragrunt-gcp-tool-mcp.git