AWS CLI

🌟 amazon-q-cli is great, and it is great because it hasuse_awsMCP tool to interact with AWS API.

💡 Wouldn't it be greater if thisuse_awswas portable, and use it across different AI tools, whichever you're currently using?

⚡use_aws_mcpis a standalone Model Context Protocol (MCP) server that provides AWS CLI functionality through a standardized interface.
This server replicates the functionality of theuse_awstool from the Amazon Q Developer CLI.

- AWS CLI Integration: Execute AWS CLI commands with proper parameter handling
- Safety Checks: Automatic detection of read-only vs. write operations
- User Agent Management: Proper AWS CLI user agent setup for tracking
- Parameter Formatting: Automatic conversion of parameters to kebab-case for CLI compatibility
- Error Handling: Comprehensive error handling and output formatting
- MCP Protocol: Full Model Context Protocol compliance
- Human-Readable Descriptions: Rich command descriptions using terminal formatting

- 🦀Rust (1.70 or later), Cargo

- for MacOS and linux, install withcurl https://sh.rustup.rs -sSf | sh

The binary will be available attarget/release/use_aws.

To use this server with an MCP client, first install it using Cargo:

{ "mcpServers": { "use_aws_mcp": { "name": "use_aws_mcp", "command": "use_aws_mcp", "timeout": 300, "env": {}, "disabled": false } } }

⚠️ Important Caveat for Using MCP Client

With q cli, mcp clients are shell process, so credentials env likeAWS_DEFAULT_PROFILEare automatically transfered to mcp server.

However, non shell mcp clients like cursor cannot take advantage of this, so it is best advised to require mcp clients directly to use specific aws profile.
- Set mcp.json above
- Set API key, or login to specific profile usingaws sso login
- Ask away mcp client aws related questions! and be sure to require it to use specific profile.

The server communicates via stdin/stdout using JSON-RPC protocol.

The server provides human-readable descriptions of AWS CLI commands. You can see this in action by running the example:

Running aws cli command: Service name: s3 Operation name: list-buckets Parameters: - max-items: "10" - query: "Buckets[].Name" Profile name: development Region: us-west-2 Label: List S3 buckets with query ✅ This command is read-only (no acceptance required)

The server provides a single tool calleduse_awswith the following schema:

{ "name": "use_aws", "description": "Execute AWS CLI commands with proper parameter handling and safety checks", "inputSchema": { "type": "object", "properties": { "service_name": { "type": "string", "description": "AWS service name (e.g., s3, ec2, lambda)" }, "operation_name": { "type": "string", "description": "AWS CLI operation name (e.g., list-buckets, describe-instances)" }, "parameters": { "type": "object", "description": "Optional parameters for the AWS CLI command", "additionalProperties": true }, "region": { "type": "string", "description": "AWS region (e.g., us-west-2, eu-west-1)" }, "profile_name": { "type": "string", "description": "Optional AWS profile name" }, "label": { "type": "string", "description": "Optional label for the operation" } }, "required": ["service_name", "operation_name", "region"] } }

{ "name": "use_aws", "arguments": { "service_name": "s3", "operation_name": "ls", "region": "us-west-2" } }

{ "name": "use_aws", "arguments": { "service_name": "ec2", "operation_name": "describe-instances", "region": "us-west-2", "parameters": { "instance-ids": "i-1234567890abcdef0" } } }

{ "name": "use_aws", "arguments": { "service_name": "lambda", "operation_name": "list-functions", "region": "us-west-2", "profile_name": "development" } }

The server automatically detects read-only operations based on the operation name prefix:

- Read-only prefixes:get,describe,list,ls,search,batch_get
- Write operations: All other operations require explicit user acceptance

Large outputs are automatically truncated to prevent memory issues, with a maximum response size of 100KB.

# Run the description demo cargo run --example description_demo

- src/lib.rs: Core library with types and constants
- src/error.rs: Error handling types
- src/use_aws.rs: Core AWS CLI functionality (replicated from original)
- src/mcp_server.rs: MCP server implementation
- src/main.rs: Binary entry point
- examples/description_demo.rs: Example demonstrating command descriptions

If you do not have Cargo (the Rust package manager) installed, you can get it by installing Rust usingrustup:

curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh

RUST_LOG=use_aws=debug ./target/release/use_aws

Agent-ready global image CDN that AI agents can install and operate through MCP.