OverviewArchitectureRecipeSourceProcessorSinkContext Graph for AI
OverviewArchitectureRecipeSourceProcessorSinkContext Graph for AI

Amazon S3

Writes records to a single object in an Amazon S3 bucket, or any S3-compatible store. Each record is one line of JSON (NDJSON).

Usage

sinks:
  - name: s3
    config:
      bucket_url: s3://bucket-name/optional-prefix
      region: us-east-1

Configuration

KeyTypeDescriptionRequired
bucket_urlstringBucket and optional prefix, e.g. s3://bucket-name/optional-prefixrequired
regionstringAWS region of the bucket, e.g. us-east-1required
object_prefixstringPrefix added to the object name, e.g. github-usersoptional
access_key_idstringAWS access key IDoptional
secret_access_keystringAWS secret access keyoptional
endpointstringCustom endpoint for S3-compatible stores like MinIO, e.g. http://localhost:9000optional

If access_key_id and secret_access_key are not set, the sink uses the default AWS credential chain (environment variables, shared config, IAM roles).

Output format

Each run creates one object named [prefix/][prefix-]<RFC3339 timestamp>.ndjson in the bucket. Every line is the whole record as JSON: {"entity": {...}, "edges": [...]} with snake_case keys. The edges field is left out when a record has no edges.

KafkaStencil
On this page
UsageConfigurationOutput format