Redpanda Connect, formerly known as Benthos, is an open-source stream processor written in Go. It connects input sources to output sinks through declarative YAML configuration files, with a pipeline stage for transformations, enrichments, and filtering. The tool handles at-least-once delivery guarantees without requiring disk-persisted state, making deployment and scaling straightforward.

A typical pipeline configuration specifies an input (such as Kafka, AWS SQS, GCP Pub/Sub, or HTTP), a list of processors that transform messages, and an output (such as Redis Streams, Elasticsearch, PostgreSQL, or S3). The built-in mapping language, Bloblang, provides a concise syntax for restructuring payloads, extracting fields, performing arithmetic, and handling errors. Pipelines can be as simple as three lines or as complex as multi-stage enrichment workflows with conditional branching.

Redpanda Connect supports a wide range of connectors: AWS services (DynamoDB, Kinesis, S3, SQS, SNS), Azure storage, GCP Pub/Sub and Cloud Storage, Kafka, NATS, MQTT, AMQP, Redis, Cassandra, MongoDB, SQL databases (MySQL, PostgreSQL, ClickHouse, MSSQL), HTTP clients and servers including WebSockets, and many more. This breadth makes it a versatile glue layer for data infrastructure.

The tool ships as a static binary and Docker image. It exposes HTTP health-check endpoints (/ping for liveness, /ready for readiness) and exports metrics for monitoring. Configuration can be provided via files, environment variable interpolation, or CLI flags. Redpanda Connect also supports running as an embedded library in Go applications for custom deployments.

With over 8,000 GitHub stars across its Benthos lineage and active development by Redpanda Data, the project is widely used in production for ETL pipelines, event-driven architectures, and real-time data integration. It is available via Homebrew, Docker, and standalone binaries.