ralph is an archived example, not installed by init. Copy it from
examples/init-pack/ with its dependency closure, or ask create-workflow
for a bounded variant. Once installed, it repeatedly runs the configured
smart agent on your prompt.

The ralph UI shows the single agent task re-running on your prompt in an unbounded loop.
Inputs
Use it when
- An open-ended prompt with no fixed finish line benefits from continuous iteration (reducing flaky tests, chipping at tech debt).
Use something else when
- The work has a clear definition of done: use
implementormission. - Need review gates or validation each pass: use a validation-loop workflow such as
implement.
Resume & retry
ralph won’t stop on its own: cancel it with bunx smthrs cancel RUN_ID when enough work is done, or configure a maxIterations limit in your workflow file. Runs are durable and resume with up --resume.