mcp/aws-api-mcp-server repository overview

Best practices for command generation:
- Always use the most specific service and operation names
- Always use the working directory when writing files, unless user explicitly mentioned another directory
- Include --region when operating across regions
- Only use filters (--filters, --query, --prefix, --pattern, etc) when necessary or user explicitly asked for it

Command restrictions:
- DO NOT use bash/zsh pipes (|) or any shell operators
- DO NOT use bash/zsh tools like grep, awk, sed, etc.
- DO NOT use shell redirection operators (>, >>, <)
- DO NOT use command substitution ($())
- DO NOT use shell variables or environment variables
- DO NOT use relative paths for reading or writing files, use absolute paths instead

Common pitfalls to avoid:
1. Missing required parameters - always include all required parameters
2. Incorrect parameter values - ensure values match expected format
3. Missing --region when operating across regions

Returns:
    CLI execution results with API response data or error message

IMPORTANT: Only use this tool when:
1. You are unsure about the exact AWS service or operation to use
2. The user's request is ambiguous or lacks specific details
3. You need to explore multiple possible approaches to solve a task
4. You want to provide options to the user for different ways to accomplish their goal

DO NOT use this tool when:
1. You are confident about the exact AWS CLI command needed - use 'call_aws' instead
2. The user's request is clear and specific about the AWS service and operation
3. You already know the exact parameters and syntax needed
4. The task requires immediate execution of a known command

Best practices for query formulation:
1. Include the user's primary goal or intent
2. Specify any relevant AWS services if mentioned
3. Include important parameters or conditions mentioned
4. Add context about the environment or constraints
5. Mention any specific requirements or preferences

CRITICAL: Query Granularity
- Each query should be granular enough to be accomplished by a single CLI command
- If the user's request requires multiple commands to complete, break it down into individual tasks
- Call this tool separately for each specific task to get the most relevant suggestions
- Example of breaking down a complex request:
  User request: "Set up a new EC2 instance with a security group and attach it to an EBS volume"
  Break down into:
  1. "Create a new security group with inbound rules for SSH and HTTP"
  2. "Create a new EBS volume with 100GB size"
  3. "Launch an EC2 instance with t2.micro instance type"
  4. "Attach the EBS volume to the EC2 instance"

Query examples:
1. "List all running EC2 instances in us-east-1 region"
2. "Get the size of my S3 bucket named 'my-backup-bucket'"
3. "List all IAM users who have AdministratorAccess policy"
4. "List all Lambda functions in my account"
5. "Create a new S3 bucket with versioning enabled and server-side encryption"
6. "Update the memory allocation of my Lambda function 'data-processor' to 1024MB"
7. "Add a new security group rule to allow inbound traffic on port 443"
8. "Tag all EC2 instances in the 'production' environment with 'Environment=prod'"
9. "Configure CloudWatch alarms for high CPU utilization on my RDS instance"

Returns:
    A list of up to 10 most likely AWS CLI commands that could accomplish the task, including:
    - The CLI command
    - Confidence score for the suggestion
    - Required parameters
    - Description of what the command does

{
  "mcpServers": {
    "aws-api": {
      "command": "docker",
      "args": [
        "run",
        "-i",
        "--rm",
        "-e",
        "AWS_REGION",
        "-e",
        "AWS_API_MCP_PROFILE_NAME",
        "-e",
        "AWS_ACCESS_KEY_ID",
        "-e",
        "AWS_SECRET_ACCESS_KEY",
        "-e",
        "AWS_SESSION_TOKEN",
        "mcp/aws-api-mcp-server"
      ],
      "env": {
        "AWS_REGION": "us-east-1",
        "AWS_API_MCP_PROFILE_NAME": "default",
        "AWS_ACCESS_KEY_ID": "AKIAIOSFODNN7EXAMPLE",
        "AWS_SECRET_ACCESS_KEY": "your-aws-secret-access-key",
        "AWS_SESSION_TOKEN": "your-aws-session-token"
      }
    }
  }
}