AWS S3 Tables MCP Server: Setup for Claude, Cursor and Other Apps

• Lists your S3 table buckets

  Returns the table buckets in your AWS account for a region.

  Try: “List my S3 table buckets.”

  Network · list_table_buckets

• Lists the namespaces in your table buckets

  Returns the namespaces, which group tables inside a table bucket.

  Try: “What namespaces do I have in S3 Tables?”

  Network · list_namespaces

• Lists your S3 tables

  Returns the tables in your table buckets.

  Try: “List all tables in the sales namespace.”

  Network · list_tables

• Runs a read-only SQL query on a table

  Runs a SQL query against S3 Tables through the Iceberg catalog. You give the table bucket ARN, region, namespace and query. It reads only.

  Try: “Find the top 3 customers by spending in the transactions table.”

  Network · query_database

• Shows where a table's metadata file lives

  Returns the S3 location of the table's metadata, which holds its schema and settings.

  Try: “Show the schema location of the customer_data table.”

  Network · get_table_metadata_location

• Shows a table's maintenance settings

  Returns the maintenance configuration of one table, such as compaction and cleanup settings.

  Try: “What maintenance is configured for the customers table?”

  Network · get_table_maintenance_config

• Shows the status of a table's maintenance jobs

  Returns whether the maintenance jobs on a table have run and how they ended.

  Try: “Did the last maintenance job on customers succeed?”

  Network · get_maintenance_job_status

• Shows the metadata table setup of a regular S3 bucket

  Returns the metadata table configuration of an ordinary S3 bucket, which feeds the S3 Metadata tables.

  Try: “Query all available metadata about test-bucket.”

  Network · get_bucket_metadata_config

• Creates a table bucket

  Creates a new S3 table bucket. Only works when the server is started with --allow-write.

  Try: “Create a table bucket named hello-world.”

  Write · create_table_bucket

• Creates a namespace in a table bucket

  Creates a namespace to group tables. Only works with --allow-write.

  Try: “Create a namespace called retail_data in hello-world.”

  Write · create_namespace

• Creates a table in a namespace

  Creates an Iceberg table with the schema you give. Only works with --allow-write.

  Try: “Create a customers table with id, name and signup_date.”

  Write · create_table

• Loads a CSV file from S3 into a table

  Reads a CSV file in S3 and imports it into an S3 table. Only works with --allow-write.

  Try: “Create an S3 table from s3://my-bucket/data.csv.”

  Write · import_csv_to_table

• Loads a Parquet file from S3 into a table

  Reads a Parquet file in S3 and imports it into an existing table. Only works with --allow-write.

  Try: “Import s3://my-bucket/events.parquet into the events table.”

  Write · import_parquet_to_table

• Adds rows to a table

  Appends the rows you give to an existing Iceberg table. It adds rows and does not update or delete. Only works with --allow-write.

  Try: “Append these 5 rows to the orders table.”

  Write · append_rows_to_table

• Renames a table or moves it to another namespace

  Renames a table, or moves it to a different namespace in the same bucket. Only works with --allow-write.

  Try: “Rename the table old_orders to orders.”

  Write · rename_table

• Points a table at a new metadata file

  Updates the location of the table's metadata file. Only works with --allow-write.

  Try: “Update the metadata location of the orders table to the new file.”

  Write · update_table_metadata_location

Watch out for

• --allow-write lets it create tables and add rows

  It is off by default. Turn it on only with an IAM policy limited to the buckets you mean, and keep a backup of important data. github.com

• Table contents go to your AI provider

  Rows returned by a query are sent to the model. Query only data you can share. github.com

Security advisories