s3_bucket_createLABSCO SUMMARY
S3 and DynamoDB, with a record of what was done.
The tool surface is wide enough for real work — buckets and objects, tables and items, batch operations and TTL — and the audit resource means the actions an agent took can be read back afterwards rather than reconstructed from the console.
What it is
An MCP server for AWS operations covering S3 and DynamoDB. Every operation it performs is logged, and the log is readable back through the audit://aws-operations resource.
What you get
- S3 buckets created, listed and deleted
- Objects uploaded, read, listed and deleted
- DynamoDB tables created, described, updated and deleted
- Items put, retrieved, updated, deleted, queried and scanned
- Batch reads, batch writes and batch statement execution
- Table TTL described and updated
- An audit record of every operation, at audit://aws-operations
Requirements
AWS credentials for an IAM user with read-write permissions on S3 and DynamoDB — either an access key, a secret and a region in the environment, or the default AWS credential chain. uv runs it.
Setup effort
Follow the published steps — Clone this repository
USE CASES
01
Getting a file into or out of a bucket
Upload it, list what is there, and read an object back — the three operations that cover most day-to-day S3 use.
02
Answering a question from a DynamoDB table
Query on the key when you have one, scan when you do not, and fetch the single item once you have found it.
03
Standing up storage for something new
Create the bucket and the table together, then describe the table to confirm the key schema is what you meant.
04
Loading or clearing a batch of records
Write and read in batches rather than one item per call, which is the difference between a seeded table and a long afternoon.
05
Setting records to expire on their own
Read the current TTL configuration and set it, so short-lived rows clean themselves up instead of accumulating.
06
Tidying up after an experiment
Empty and remove the bucket, and drop the table, so a trial does not quietly become a line on the bill.
22 tools
Listed in project docs. Your client may show a different list.
s3_bucket_lists3_bucket_deletes3_object_uploads3_object_deletes3_object_lists3_object_readdynamodb_table_createdynamodb_table_describedynamodb_table_deletedynamodb_table_updatedynamodb_item_putdynamodb_item_getdynamodb_item_updatedynamodb_item_deletedynamodb_item_querydynamodb_item_scandynamodb_batch_getdynamodb_item_batch_writedynamodb_batch_executedynamodb_describe_ttldynamodb_update_ttlBEFORE YOU INSTALL
What to know first
Tools you get
22 tools named in the project's documentation. Despite the name, it supports S3 and DynamoDB only. Every operation is logged to the audit://aws-operations resource.
Requires
AWS credentials for an IAM user with read-write access to S3 and DynamoDB — AWS_ACCESS_KEY_ID and AWS_SECRET_ACCESS_KEY (with AWS_REGION, default us-east-1) in your environment, or the default AWS credential chain set up with aws configure
Where it runs
The documented client runs the server on your machine.
INSTALLATION
Connect your client
Server setup
Clone this repository
Set up AWS credentials either via the env vars AWS_ACCESS_KEY_ID / AWS_SECRET_ACCESS_KEY / AWS_REGION, or via the default AWS credential chain (aws configure) — the server needs an IAM user with RW permissions for S3 and DynamoDB
Add the config block below to claude_desktop_config.json (macOS: ~/Library/Application Support/Claude/claude_desktop_config.json, Windows: %APPDATA%/Claude/claude_desktop_config.json)
Install and open the Claude desktop app
MCP Clients
Claude Desktop
CLAUDE DESKTOP · JSON
{
"mcpServers": {
"mcp-server-aws": {
"command": "uv",
"args": [
"--directory",
"<YOUR_PATH_TO_REPO_MCP_SERVER_AWS>",
"run",
"mcp-server-aws"
]
}
}
}The configuration on this page does not set AWS_ACCESS_KEY_ID or AWS_SECRET_ACCESS_KEY. The README gives two ways: set them (with AWS_REGION, default us-east-1) as environment variables, or use the default AWS credential chain set up with aws configure. The IAM user needs read-write access to S3 and DynamoDB.
