AWSSAFE
A Model Context Protocol server implementation for operations on AWS resources
Overview
From the repository's own README, as read at the audited commit. Badges and raw HTML are left out.
[](https://smithery.ai/server/mcp-server-aws)
A Model Context Protocol server implementation for AWS operations that currently supports S3 and DynamoDB services. All operations are automatically logged and can be accessed through the audit://aws-operations resource endpoint.
See a demo video here.
Listed as a Community Server within the MCP servers repository.
Running locally with the Claude desktop app
Installing via Smithery
To install AWS MCP Server for Claude Desktop automatically via Smithery:
npx -y @smithery/cli install mcp-server-aws --client claude
Manual Installation
- Clone this repository.
- Set up your AWS credentials via one of the two methods below. Note that this server requires an IAM user with RW permissions for your AWS account for S3 and DynamoDB.
- Environment variables:
AWS_ACCESS_KEY_ID,AWS_SECRET_ACCESS_KEY,AWS_REGION(defaults tous-east-1) - Default AWS credential chain (set up via AWS CLI with
aws configure) - Add the following to your
claude_desktop_config.jsonfile: - On MacOS:
~/Library/Application\ Support/Claude/claude_desktop_config.json - On Windows:
%APPDATA%/Claude/claude_desktop_config.json
"mcpServers": {
"mcp-server-aws": {
"command": "uv",
"args": [
"--directory",
"/path/to/repo/mcp-server-aws",
"run",
"mcp-server-aws"
]
}
}- Install and open the [Claude desktop app](https://claude.ai/downl
69c51c2319b9OBSERVED · 2026-10-07Connect
Built from this server's own package name, version and transport as found in its source — not copied from anyone's documentation, so it cannot drift against a page we do not control. Replace the environment placeholders with a token scoped to the least it needs.
claude mcp add mcp-server-aws --env AWS_ACCESS_KEY_ID=${AWS_ACCESS_KEY_ID} --env AWS_SECRET_ACCESS_KEY=${AWS_SECRET_ACCESS_KEY} -- uvx mcp-server-aws{
"mcpServers": {
"mcp-server-aws": {
"command": "uvx",
"args": [
"mcp-server-aws"
],
"env": {
"AWS_ACCESS_KEY_ID": "${AWS_ACCESS_KEY_ID}",
"AWS_SECRET_ACCESS_KEY": "${AWS_SECRET_ACCESS_KEY}"
}
}
}
}Exposed tools (23)
10 read · 8 write · 5 destructive. Blast radius: 5 tools can delete or overwrite — an agent that can be talked into calling a tool can be talked into calling this one.
| Tool | Risk | Description |
|---|---|---|
dynamodb_batch_execute | write | Execute multiple PartiQL statements in a batch |
dynamodb_batch_get | read | Batch get multiple items from DynamoDB tables |
dynamodb_describe_ttl | read | Get the TTL settings for a table |
dynamodb_item_batch_write | destructive | Batch write operations (put/delete) for DynamoDB items |
dynamodb_item_delete | destructive | Delete an item from a DynamoDB table |
dynamodb_item_get | read | Get an item from a DynamoDB table |
dynamodb_item_put | write | Put an item into a DynamoDB table |
dynamodb_item_query | read | Query items in a DynamoDB table |
dynamodb_item_scan | read | Scan items in a DynamoDB table |
dynamodb_item_update | write | Update an item in a DynamoDB table |
dynamodb_table_create | write | Create a new DynamoDB table |
dynamodb_table_delete | destructive | Delete a DynamoDB table |
dynamodb_table_describe | read | Get details about a DynamoDB table |
dynamodb_table_list | read | List all DynamoDB tables |
dynamodb_table_update | write | Update a DynamoDB table |
dynamodb_update_ttl | write | Update the TTL settings for a table |
s3_bucket_create | write | Create a new S3 bucket |
s3_bucket_delete | destructive | Delete an S3 bucket |
s3_bucket_list | read | List all S3 buckets |
s3_object_delete | destructive | Delete an object from S3 |
s3_object_list | read | List objects in an S3 bucket |
s3_object_read | read | Read an object |
s3_object_upload | write | Upload an object to S3 |
Trust audit
SAFEgrade B · trust 89/100 Nothing in the source contradicts what it says it does. Grade A is reserved for packages that have also passed the behavioural sandbox.
| Layer | What it checks | Result |
|---|---|---|
| L0 | Provenance & inventory | PASS |
| L1 | Static analysis of the code | PASS |
| L2 | Instruction surface (what it tells the agent) | PASS |
| L3 | Class-specific surface | WARN |
| L4 | Behavioural (sandbox) | SKIPPED |
What the source does
- Filesystem
- declared (1 observation(s))
- Network
- none-observed
- Shell
- none-observed
- Dependencies
- pinned
- Secrets in source
- none-found
Findings (2)
dynamodb_item_batch_write, dynamodb_item_delete, dynamodb_table_delete, s3_bucket_delete, s3_object_delete
io.BytesIO(base64.b64decode(arguments["file_content"])),
Gates applied: no_behavioural_pass.
69c51c2319b9full audit observations/trust-audit/mcp-server/rishikavikondala__aws-2.json · Report an issue / request a re-scanAudit history
Every audit this server has had. A grade with a past is a grade somebody is still checking.
| Date | Source | Verdict | Grade | Score | Change |
|---|---|---|---|---|---|
| 2026-10-07 | 69c51c2319b9 | SAFE | B | 89 | first audit |
Questions
What is the AWS MCP server?
A Model Context Protocol server implementation for operations on AWS resources
What tools does AWS expose?
23 in total: 10 read-only, 8 that write, and 5 that can delete or overwrite (dynamodb_item_batch_write, dynamodb_item_delete, dynamodb_table_delete, s3_bucket_delete, s3_object_delete). Every one is listed on this page with its risk.
Is AWS safe to connect to an agent?
The audit found nothing in the source that contradicts what it says it does, and graded it B (89/100). Grade A is held back for packages that have also passed a sandboxed behavioural run, which is why a clean server reads B. Separately from the audit: 5 of its tools can destroy data, so scope the token you give it to what you actually need.
What credentials does AWS need?
It reads AWS_ACCESS_KEY_ID and AWS_SECRET_ACCESS_KEY from the environment. Give it a token scoped to the least it needs — an agent that can be talked into calling a tool can be talked into calling it with your credentials.
How does AWS run?
It speaks stdio, so it runs as a local process your client starts. It is published on PyPI as mcp-server-aws.
How current is this page?
The grade is for one exact copy of the source (69c51c2319b9), read on 2026-10-07. The repository is watched and re-audited when it changes.