AWS CDK MCP Server

Overview

AWS CDK MCP Server is a stdio MCP server for AWS Cloud Development Kit guidance, infrastructure-as-code patterns, CDK Nag security/compliance guidance, AWS Solutions Constructs discovery, GenAI CDK construct discovery, Lambda layer documentation, and Bedrock Agent schema generation.

The upstream README marks this server deprecated and recommends AWS IaC MCP Server instead. This submission reviews the published PyPI package awslabs.cdk-mcp-server version 1.0.15. The current GitHub subfolder and server-specific generated docs URL returned 404 during review, so the official PyPI source distribution is the authoritative reviewed source for package metadata, source files, and documentation.

Install and launch

The documented default MCP client launch uses uvx:

{
  "command": "uvx",
  "args": ["awslabs.cdk-mcp-server@latest"],
  "env": {
    "FASTMCP_LOG_LEVEL": "ERROR"
  }
}

The README also documents an alternate Windows uv tool run --from awslabs.cdk-mcp-server@latest awslabs.cdk-mcp-server.exe example and optional Docker build/run commands. Those alternate command flags are install or alternate-client documentation only; they are not package arguments for the default uvx target.

Configuration

FASTMCP_LOG_LEVEL is the only environment variable in the documented default uvx launch. AWS_PROFILE and AWS_REGION appear in the alternate Windows client example only and were not found as direct server source reads, so they are excluded from the default package transport environment. ENABLE_CDK_NAG appears in CDK application guidance as an example user-app setting and is not read by the MCP server runtime. Dockerfile variables are image-internal settings and are recorded in source review rather than default package launch configuration.

Tools

• CDKGeneralGuidance: returns prescriptive CDK best-practice guidance.
• GetAwsSolutionsConstructPattern: searches or retrieves AWS Solutions Constructs patterns by pattern name or AWS service list.
• SearchGenAICDKConstructs: searches GenAI CDK constructs by query and/or construct type.
• GenerateBedrockAgentSchema: generates OpenAPI schemas from local Lambda Python files using BedrockAgentResolver and writes output schema/fallback files to client-provided paths.
• LambdaLayerDocumentationProvider: provides Lambda layer documentation and code-generation guidance.
• ExplainCDKNagRule: explains a CDK Nag rule and remediation guidance.
• CheckCDKNagSuppressions: checks CDK code or a local file path for CDK Nag suppressions.

Resources

Registered resource URI families include cdk-nag://rules/{rule_pack}, cdk-nag://warnings/{rule_pack}, cdk-nag://errors/{rule_pack}, aws-solutions-constructs://{pattern_name}, genai-cdk-constructs://..., and lambda-powertools://{topic}.

Prerequisites

Requires uv and Python 3.10+. The README documents AWS CDK CLI installation with npm install -g aws-cdk for the CDK workflows the server guides; that implies Node.js/npm for CDK CLI installation. Some tools fetch documentation dynamically from GitHub or AWS documentation. Bedrock Agent schema generation requires Lambda code using AWS Lambda Powertools BedrockAgentResolver; missing Lambda dependencies may cause the tool to create a fallback script that must be run in a properly configured local environment.

Limitations and security notes

The server runs over stdio. External documentation features depend on GitHub/AWS documentation availability and structure. File-oriented tools operate on client-provided paths: suppression checking can read local code files, and schema generation can read Lambda files, create simplified variants, write OpenAPI schemas, and write fallback scripts. CDK Nag suppressions require human review; users should review warnings and avoid suppressing issues whenever possible.