finds.dev← search

// the find

runabol/tork

★ 820 · Go · MIT · updated Jul 2026

Tork is a lightweight, distributed workflow engine that runs tasks as simple scripts within Docker containers.

Tork is a Go workflow engine that runs each task as a script inside a Docker, Podman, or shell runtime. A coordinator schedules jobs defined in YAML, and workers execute the tasks. It suits teams that want a self-hosted pipeline runner and can run PostgreSQL and a message broker to get it. The description calls it 'lightweight', which holds for standalone mode but not for distributed mode.

- With the Docker or Podman runtime, each task gets its own container with CPU and memory limits, a timeout, and up to 10 retries, all declared in the job file. A failing task stays contained to that task.

- The coordinator is stateless and leaderless, with state in Postgres and messages moving over a broker, so you scale by adding workers. The engine is also a Go library with registration hooks for custom brokers, datastores, mounters, and middleware, so you can embed it instead of forking it.

- Fan-out and nesting live in the job format: parallel blocks, each loops with a concurrency limit, sub-jobs with a detached mode, and pre/post tasks that share the main task's mounts. You get a reasonable set of workflow primitives without a separate DAG tool.

- The README says secrets are redacted and then tells you to avoid printing them. As far as the docs describe it, redaction is a filter on logs, so a secret the task transforms before printing, such as base64-encoded or split across lines, will probably get through.

- The Shell runtime runs tasks as plain host processes. The uid and gid settings are empty by default, so an unconfigured Shell runtime has no isolation. The README gives this one line of caution.

- Distributed mode needs PostgreSQL, RabbitMQ, and a Docker or Podman daemon on every worker before you get anything past standalone. Podman is listed under 'runtime agnostic', but GPU support and the Docker config file are documented only for Docker, so it is unclear how much of the feature set Podman covers.

View on GitHub →

// want more like this?

We dig through GitHub every week and send a few repos picked for what you actually care about — each with an honest take like this one.

Get finds in your inbox → Search again →