Skip to main content

cdk-agent-mcp-server

MCP server that generates, validates, and deploys AWS CDK infrastructure from natural language.

Delegates to AWS MCP servers for validation, documentation, and pricing — no duplicate tooling.

Architecture

cdk-agent-mcp-server (stdio)
  ├── clarify_requirements    → Bedrock (requirements gathering before design)
  ├── design_architecture     → AWS IaC MCP (search_cdk_documentation)
  ├── confirm_plan            → Bedrock (human/agent approval gate)
  ├── generate_cdk_stack      → AWS IaC MCP (search_cdk_documentation + search_cdk_samples_and_constructs)
  ├── validate_and_repair     → real cdk synth + AWS IaC MCP (validate_cloudformation_template + check_cloudformation_template_compliance)
  ├── estimate_cost           → AWS Pricing MCP (get_pricing + generate_cost_report)
  ├── lookup_cdk_construct    → AWS IaC MCP (search_cdk_documentation + search_cdk_samples_and_constructs + cdk_best_practices)
  ├── troubleshoot_deployment → AWS IaC MCP (troubleshoot_cloudformation_deployment)
  ├── deploy_stack            → real cdk deploy (requires approved=true)
  ├── destroy_stack           → CloudFormation DeleteStack (requires approved=true)
  ├── generate_diagram        → local Mermaid generation
  └── full_pipeline           → orchestrates all of the above

HITL (Human-in-the-Loop) Flow

User/Agent describes intent
       │
       ▼
clarify_requirements  →  returns structured questions + defaults
       │
       ▼
User/Agent answers (or accepts defaults)
       │
       ▼
design_architecture   →  generates plan using enriched context
       │
       ▼
confirm_plan          →  summarizes plan for approval (services, cost tier, security)
       │
       ▼
User/Agent approves (or modifies)
       │
       ▼
generate_cdk_stack → validate_and_repair → estimate_cost → deploy_stack

Prerequisites

  • Python 3.10+, Node.js 18+ (for cdk synth/deploy), uv installed
  • AWS CLI configured, Bedrock Claude Sonnet 4 enabled

Install

{
  "mcpServers": {
    "ict-mcp-server": {
      "command": "uvx",
      "args": ["ict-mcp-server"],
      "env": { "AWS_REGION": "us-east-1" }
    }
  }
}

Works with Kiro (~/.kiro/settings/mcp.json), Cursor (.cursor/mcp.json), Claude Code (claude mcp add), Amazon Quick Desktop (Settings → Capabilities → MCP).

Agent

Task Delegates To What It Does
clarify_requirements Bedrock + Knowledge MCP NL → structured clarifying questions with defaults
full_pipeline All below End-to-end: describe → clarify → plan → confirm → diagram → CDK → validate → costs
design_architecture IaC MCP search_cdk_documentation NL + requirements → architecture plan JSON
confirm_plan Bedrock Plan → human-readable summary for approval
generate_cdk_stack IaC MCP search_cdk_documentation + search_cdk_samples_and_constructs Plan → CDK TypeScript
validate_and_repair cdk synth + IaC MCP validate_cloudformation_template + check_cloudformation_template_compliance Validate + auto-repair loop
estimate_cost Pricing MCP get_pricing + generate_cost_report Real AWS pricing data
lookup_cdk_construct IaC MCP search_cdk_documentation + search_cdk_samples_and_constructs + cdk_best_practices CDK docs + samples + best practices
deploy_stack cdk deploy Real deployment (requires approved=true)
destroy_stack CloudFormation DeleteStack Delete stack (requires approved=true)
troubleshoot_deployment IaC MCP troubleshoot_cloudformation_deployment Root cause analysis for failed deploys
generate_diagram Local Architecture plan → Mermaid diagram

What We Don't Reimplement

Capability Provided By
cfn-lint (template validation) AWS IaC MCP Server
cfn-guard (compliance checking) AWS IaC MCP Server
CDK documentation search AWS IaC MCP Server
CDK code samples AWS IaC MCP Server
CDK best practices AWS IaC MCP Server
Deployment troubleshooting AWS IaC MCP Server
AWS service pricing AWS Pricing MCP Server
Cost report generation AWS Pricing MCP Server

Our server only implements: architecture planning (Bedrock), CDK code generation (Bedrock), Mermaid diagrams (local), cdk synth/deploy (subprocess), and the orchestration glue.

Metadata

Release files for ict-mcp-server 0.5.1

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for ict-mcp-server 0.5.1
File Size Uploaded Provenance Actions
ict_mcp_server-0.5.1.tar.gz 61.1 kB Without provenance Details for ict_mcp_server-0.5.1.tar.gz

Built distribution (wheel)

Table of built distributions (wheels) for ict-mcp-server 0.5.1
File Interpreter ABI Platform Actions
ict_mcp_server-0.5.1-py3-none-any.whl Python 3 none any Details for ict_mcp_server-0.5.1-py3-none-any.whl

Total release size: 72.4 kB

Release files / ict_mcp_server-0.5.1.tar.gz

Download URL ict_mcp_server-0.5.1.tar.gz
Size 61.1 kB
Tags Source
SHA-256 checksum
How to use checksums
b3f5ebd0d71f2e29fab63c84b1616e1a132079dc6d0a169e46e47fc86ce79699
BLAKE2b-256 checksum
How to use checksums
fa2f72dfd3f3e221665200f1b59f187959e2154a0a218b80e6fab5116d6beeb0
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/6.2.0 CPython/3.13.11

Release files / ict_mcp_server-0.5.1-py3-none-any.whl

Download URL ict_mcp_server-0.5.1-py3-none-any.whl
Size 11.3 kB
Tags Python 3
SHA-256 checksum
How to use checksums
95a1619a8f392479e2e993ba81f2b4ca5ddf3d5c6ea6a84979931f3145260bb8
BLAKE2b-256 checksum
How to use checksums
0fb44d13ee81ce82bb06636792fb25ff8810f08dd5c4e104c1004e1a1ebf91c7
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/6.2.0 CPython/3.13.11

Security for ict-mcp-server 0.5.1

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

No provenance found

None of the files in this release were uploaded with provenance metadata.

A note on package security

PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.

To learn more about protecting yourself from malicious actors, visit our package security documentation.

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page