Back to directory

Workhorse

A durable task queue implemented as versioned SQL functions inside PostgreSQL, giving TypeScript, Python and Go workers the same queue, durable waits and dashboard over one SQL protocol without a separate broker.

Workhorse

Workhorse is a task queue built as versioned SQL functions inside the PostgreSQL database an application already runs, so every worker language gets the same queue behavior, the same durable waits and the same dashboard with no broker or server besides PostgreSQL. Tasks are enqueued in the same transaction as the application's database change, so a rollback cannot leave one without the other.

PostgreSQL owns queue state, scheduling, retries, recovery and immutable history. Workers claim tasks with fence tokens, so a worker that resumes after losing its lease cannot overwrite newer work, and handlers get at-least-once delivery, so external side effects still need their own idempotency keys or compensation. A dashboard shows task activity, lifecycle events, system health, worker capacity, queues and schedules without exposing the underlying tables.

Built with a TypeScript package, @stablemates/workhorse, with schema-installation tooling, plus Python and Go workers on the same SQL protocol. Requires Node.js 22 or 24 and PostgreSQL 15 through 18. Public beta; Apache-2.0 licensed.

More in Backend

See all 89
  • The Pipeline Framework

    pipelineframework.org

    A Java framework for strongly typed application flows that combine model decisions, authorized SaaS capabilities and deterministic business logic, generating the surrounding transport, persistence, retries, durable effects and observability.

  • A reference OAuth2 authorization server built with Spring Boot 4 and Java, wired up with MariaDB, Redis, OpenTelemetry tracing, and optional AI advisor integrations through Ollama, for developers to run and test locally.

  • Vyral

    openvyral.com

    An open-source, local-first contract layer and runtime for applications that need canonical data storage, retrieval and RAG, and durable execution, without locking application code to one cloud provider.