Skip to main content

MCP Server Overview

Cloudaware MCP Server provides a secure AI data layer on top of your Cloudaware CMDB and Google BigQuery Export Project.

MCP Server enables LLM-based tools such as ChatGPT, Claude, GitHub Copilot, and other AI assistants to:

  • Understand the Cloudaware CMDB as a structured data model.
  • Explore object types, fields, and relationships.
  • Execute SQL queries against your BigQuery export.

This allows AI assistants to navigate infrastructure data, answer questions about resources and dependencies, and generate code and reports based on real Cloudaware data.

Tools​

The server exposes a set of tools via a JSON-RPC interface for metadata discovery and data querying. These tools are designed to be used step by step:

ToolDescription
search_typesSearches for object types by keywords to identify API names and table IDs.
search_fieldsRetrieves field metadata for a specific type, including labels, types, and descriptions.
get_relationship_graphGets a graph of relationships and join paths between objects.
analyze_fieldGets extensive data about the values stored in a specific field.
execute_queryExecutes a BigQuery SQL query against datasets in your Export Project.

MCP clients retrieve the full tool list automatically when they connect. To get detailed tool descriptions, inputs, and outputs yourself, send a tools/list request to the server:

{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "tools/list"
}

AI Processing Flow​

A typical AI agent connected to MCP follows this sequence:

  1. Discover schema. Use search_types, search_fields, and get_relationship_graph tools.
  2. Plan queries. Understand joins, fields, and object structure. Use the analyze_field tool to check the values a field holds.
  3. Generate SQL. Build correct queries using discovered metadata.
  4. Execute SQL. Run queries through the execute_query tool.
  5. Interpret results. Produce summaries, resource lists, or code.

Key Capabilities​

CMDB as a Self-Describing Graph​

MCP exposes the Cloudaware CMDB as a self-describing graph of objects, fields, and relationships. Any compatible LLM can:

  • Discover object types and their API names.
  • Inspect fields, labels, and data types.
  • Understand how objects are related in the relationship graph.
  • Reason about infrastructure without navigating the Cloudaware UI.

Direct SQL Access to BigQuery Export​

MCP provides AI tools with controlled SQL execution against your Cloudaware BigQuery Export Project. AI assistants can:

  • Explore datasets and tables.
  • Join infrastructure and cost or usage data.
  • Build custom reports and dashboards.
  • Support engineering and analytics workflows.

Queries are executed with the execute_query tool against datasets in your existing Export Project, with the SObjects Dataset as the primary source.

Bring Your Own LLM​

MCP is model-agnostic and integrates with:

  • ChatGPT
  • Claude Code and Claude.ai
  • GitHub Copilot
  • Gemini CLI
  • LM Studio and local models
  • Internal chatbots and AI assistants
  • Other MCP-enabled tools

Your teams can keep using their existing IDEs, CLIs, and internal assistants while MCP makes Cloudaware data available within those tools.

Enterprise-Grade Security​

MCP is designed as a read-only execution layer operating under the authenticated identity.

  • AI can read metadata and execute queries only within that identity's permissions. MCP does not introduce additional privileges.
  • Access is scoped to the selected Export Project and SObjects Dataset.
  • With Google OAuth 2.0, queries run under the signed-in Google user's own BigQuery permissions. Local runs use your gcloud credentials.
  • All access respects existing BigQuery IAM permissions.

Use Cases​

AI Resource Finder​

MCP enables AI-driven resource discovery without manual UI searches. Engineers, FinOps analysts, and architects can ask an AI assistant to:

  • Find all EBS volumes matching a specific pattern.
  • List S3 buckets in selected accounts with their key properties.
  • Show all resources that match defined filters.

AI Navigator for CMDB​

MCP makes the CMDB self-explanatory for onboarding, troubleshooting, and in-product AI assistance. Users can ask an AI assistant to:

  • Explain what a specific object or field represents.
  • Show where a given CI is stored.
  • Find objects related to a service, account, VPC, or environment.

This is the backend capability for any AI helper embedded into the Cloudaware UI.

Inventory Research and Data Exploration​

With MCP, an LLM can explore the CMDB structure, explain objects and values, and guide engineers through the data without requiring them to first build complex list views or reports.

Example prompts include:

  • Analyze inventory and show what exists for a specific application.
  • Compare sets of resources under given conditions.
  • Suggest where to look for a CI or attribute.

Developer Automation Accelerator​

Before MCP, developers typically needed to:

  • Study CMDB sObjects and their relationships.
  • Build ETL jobs or many API calls.
  • Manually discover which fields to use in scripts.

With MCP, an AI assistant can:

  • Generate Python scripts that interact with Cloudaware APIs or BigQuery.
  • Select the correct objects and fields based on metadata.
  • Compose queries and joins that follow the actual relationship graph.

This reduces the time needed to build automation from weeks or months of discovery to minutes of interaction with an AI tool.

Auto-Documentation and CMDB Intelligence​

MCP enables AI-driven documentation of your data model. AI assistants can:

  • Document the structure of CMDB objects.
  • Describe field purpose and semantics.
  • Build dependency graphs between services and resources.

This is especially valuable for complex, long-running environments and for onboarding new team members.

Refer to Configuration for prerequisites, connection methods, authentication, and setup instructions.

On this page