Skip to main content

​
What is MCP?

The Model Context Protocol (MCP) is a standard for providing context to AI models. Anyshift’s MCP server gives AI coding assistants real-time access to your infrastructure knowledge graph, so they can understand your AWS, Terraform, Kubernetes, and monitoring setup. For more about MCP, visit Anthropic’s MCP documentation.

​
Why Use Anyshift MCP?

When working with infrastructure in your IDE: Without Anyshift MCP:
  • You manually look up resource IDs in the AWS console
  • You check Terraform files for dependencies
  • You search documentation for configuration details
  • You copy-paste between tools
With Anyshift MCP:
  • Your AI assistant knows your actual infrastructure
  • It references real resource IDs and configurations
  • It understands dependencies between resources
  • It generates accurate code for your environment

​
Example Workflows

​
Find and Reference Resources

You: "Update the security group for our prod database to allow 
     traffic from the new microservice"

Claude: Uses Anyshift to find:
        - The actual security group ID (sg-abc123)
        - Current ingress rules
        - The new microservice's security group

Claude generates accurate Terraform with real resource references.

​
Understand Dependencies

You: "What would break if I delete this IAM role?"

Claude: Queries Anyshift and responds:
        "This IAM role is used by:
        - 3 Lambda functions (list them)
        - 2 EC2 instances via instance profile
        - 1 ECS task definition
        Deleting it would break all these services."

​
Investigate Issues

You: "Why can't my Lambda reach the RDS database?"

Claude: Traces through your infrastructure:
        "The Lambda's security group doesn't have an outbound 
        rule for port 5432, and the RDS security group doesn't 
        allow inbound from the Lambda's security group."

​
Track Changes

You: "What changed in our infrastructure yesterday?"

Claude: Queries your 7-day history and summarizes:
        "3 resources were modified:
        - Security group sg-xyz: new rule added
        - RDS instance: storage increased
        - IAM policy: permissions updated"

​
Supported AI Tools

Cursor

IDE-integrated AI assistant for infrastructure coding

Claude Code

Terminal-based AI for infrastructure automation

Claude Desktop

Desktop AI assistant for infrastructure queries

Smithery

Access via MCP marketplace

​
Quick Setup

​
1. Generate Your API Token

  1. Go to app.anyshift.io/integrations
  2. Navigate to the MCP section
  3. Click “Generate Token”
  4. Copy the generated token

​
2. Configure Your AI Tool

Add to your MCP configuration: 
{
  "mcpServers": {
    "anyshift-mcp-server": {
      "command": "npx",
      "args": ["-y", "@anyshift/anyshift-mcp-server"],
      "env": {
        "API_TOKEN": "your-token-here"
      }
    }
  }
}
See the specific setup guides for Cursor, Claude Desktop, or Claude Code.

​
What Your AI Assistant Can Access

Once connected, your AI assistant can:
CapabilityExample Questions
Find resources”What EC2 instances are in prod?”
Map dependencies”What depends on this VPC?”
Check configurations”What are the security group rules for prod-db?”
Track changes”What changed in the last 24 hours?”
Compare IaC vs Cloud”What resources exist in AWS but not in Terraform?”
Query history”What did this security group look like yesterday?”

​
Get Started

Create Account

Sign up for Anyshift

Request Demo

See MCP integration in action