Loop runners loops
The built-ins, plugins and harnesses that keep an agent looping until the job is done: Claude Code /goal and /loop, the Ralph loop, Codex Goal Mode and the rest.
Keeps Claude working until a condition you can verify is true. The built-in version of the Ralph loop, no script needed.
Runs a task on a cadence you describe in plain English, like a scheduler inside your session.
The official Ralph loop: a stop hook feeds your prompt back in every time Claude tries to finish, until the task is done.
The original pattern: a bash loop that starts a fresh agent session on the same prompt each time, with progress kept in files.
Codex Goal ModeBuilt-in
Codex's built-in long-horizon mode: it keeps working toward an objective you set until it is met.
A skill that finds, adapts, creates and runs loops for you from the Loop Library, and audits loops that misbehave.
SuperpowersPlugin
A popular skills plugin built around loops: brainstorm, plan, test-driven development and subagent-driven builds.
Scheduled tasks in ClaudeBuilt-in
Recurring workflows that run on a schedule with your files, CRM, email and calendar connected. Loops for non-coders.
GitHub Agentic WorkflowsBuilt-in
Agent loops defined in markdown that run on a schedule or trigger in GitHub Actions: triage, CI fixes, reports.
Ralph PRD loopHarness
Ralph pointed at a PRD: works through the task list one story per iteration, marking each done.
ralph-claude-codeHarness
A Ralph runner with smarter exit detection, so the loop ends when the work is really finished.
Continuous-autonomy loops for agents you build yourself on the Vercel AI SDK.
ralph-loop (sandboxed)Harness
Runs long Ralph loops in parallel Docker sandboxes, so several can work at once safely.
Hermes AgentHarness
A self-improving CLI agent with memory that writes its own skills as it goes.
The Ralph PlaybookGuide
The principles behind loops that work: small context, backpressure from tests and types, and human oversight.