looper

Design agent loops before you run them. Goal -> Plan -> Review -> Deliver -> Judge -> Stop clean.

Quick start | Example loop | How it works

Example loop diagram

Looper turns a fuzzy automation idea into a reviewable loop shape before any runner starts changing files. This example comes from examples/ai-workflow-mapping.

flowchart TD
    G["Goal + context<br/>process notes + definition of done"] --> P["Draft plan.md<br/>host: codex / gpt-5"]
    P --> PG{"Plan gate<br/>judge: reviewer-1"}
    PG -- "revise <= 3" --> P
    PG -- "pass" --> D["Write delivery-N.md<br/>map the workflow"]
    D --> DG{"Delivery gate<br/>programmatic check + judge"}
    DG -- "revise <= 3" --> D
    DG -- "pass" --> F["Final output<br/>all gates clean"]

    S["State + log<br/>state.json + run-log.md"] -. "records" .-> P
    S -. "records" .-> D
    Stop["Stop guards<br/>max 12 iterations<br/>no progress x2<br/>budget caps"] -. "watch" .-> PG
    Stop -. "watch" .-> DG

A loop design coach for Claude Code. Looper is a skill that helps you design a good agent loop — a sharp goal, checkable verification, and a second model in the review seat — then lets you run it in the same session or save it as a portable spec. It is a design layer first: it writes files and hands the current session a clear execution prompt.

Invoke it with /looper. It interviews you, critiques your design against built-in best-practice rubrics, lets you wire in a cross-model reviewer or judge (including non-Claude models), shows you the loop as a terminal-friendly ASCII flow preview, and writes out RUN_IN_SESSION.md, loop.yaml, a compiled loop.resolved.json, a human-readable LOOP.md, a thin run-loop.py you own and edit, plus an empty loop-workspace/ and a README for the loop.

Maintainer: Kevin Simback · GitHub @ksimback · X @ksimback License: MIT

---

Where Looper fits among Claude Code's loops

The Claude Code team's own taxonomy ("Getting started with loops") sorts loops by what you hand off: turn-based loops hand off the check (verification skills), goal-based loops hand off the stop condition (/goal), time-based loops hand off the trigger (/loop, /schedule), and proactive loops hand off the whole prompt (routines composing all of the above). Every one of those primitives runs a loop. Looper is the layer in front of them: it helps you design a loop that's worth running, then emits a spec any of the four can execute.

Concretely, what each loop type asks you to hand off is exactly what Looper coaches and hardens:

• Turn-based — the check you'd encode as a verification skill is Looper's typed verification block: programmatic first, judge rubric second, human signoff last.
• Goal-based — /goal's stop condition is Looper's definition_of_done plus gates; the difference is who judges it (a model family you chose, against a typed rubric) and what surrounds it (revision caps, no-progress stalls, budget guards).
• Time-based — /loop and /schedule re-fire whatever you give them; give them a loop that already passed compile and lint, and each firing follows RUN_IN_SESSION.md or run-loop.py instead of an ad-hoc prompt.
• Proactive — a routine is only as good as the prompt it repeats; the compiled loop.resolved.json is a versionable, reviewable artifact you can hand to a routine and audit later.

What /goal actually does

/goal sets a persistent objective for the session. Once set, Claude keeps it as a reference point, checks after each significant action whether the current state satisfies the goal, and keeps working until it does — so it doesn't stop and ask after every step.

That's genuinely useful for persistence. But three things are missing for serious work:

• No coaching. /goal takes whatever goal you type, however vague. It won't tell you the goal is unfalsifiable or that "done" was never defined. Garbage goal in, confidently-wrong loop out.
• Single-vendor evaluation. The stop condition is checked by an evaluator model, but it's the same vendor in the same pipeline — not a model family you chose for blind-spot coverage, and there's no typed rubric behind the verdict. A review council exists precisely to put a different set of eyes, with explicit criteria, on the work.
• No structure to inspect or reuse. The goal lives in the session, not as a portable, versionable artifact. There's no typed verification, no explicit gates, no second model.

What /loop and /schedule actually do

/loop is a scheduler. You give it an interval and a task; it re-fires the prompt or skill on that cadence — polling CI, watching a deploy, monitoring a background job. (Omit the interval and it self-paces.) /schedule is the same hand-off moved to the cloud: a routine that keeps firing when your machine is off.

They're the right tool for "run this thing every five minutes until I say stop." Neither is a loop designer: they don't help you decide what runs, define success criteria, or bring in a reviewer. They schedule; they don't critique.

Side by side

Looper is the design layer that sits in front of all of them. It produces a well-specified loop — coached goal, typed verification, a cross-model gate — then gives you a default in-session handoff prompt plus a portable spec. The same design can be run immediately in the conversation, driven by /goal for persistence, fired on a schedule by /loop or /schedule, or run later with Python. Looper doesn't replace them; it gives them something good to run.

	/goal	/loop / /schedule	Looper
Layer	execution (in-session)	execution (scheduling, local / cloud)	design (pre-flight)
You hand off	the stop condition	the trigger	the design, checked before anything runs
Coaches your goal	no	no	yes
Typed, checkable verification	no	no	yes (programmatic / judge / human)
Reviewer model	built-in evaluator, same vendor	none	a different model family, by default
Explicit review gates	implicit	none	plan gate + delivery gate
Termination guards	goal-condition only	interval / until	iteration + revision + no-progress + budget caps
Portable, versionable artifact	no	the cron job / routine	loop.yaml + resolved spec
Static design checks	no	no	looper lint (CI-friendly)
Runs the loop	yes	yes	yes, by handing the current session a runnable prompt; Python runner optional

The honest summary: if you already know your loop is well-designed and you just need it to persist or to fire on a schedule, /goal, /loop, and /schedule are the right reach. Looper exists for the part those don't touch — making sure the loop is worth persisting before you hand it off, and making sure something other than the author is checking the work.

Sources for the behavior described above: the Claude Code team's "Getting started with loops" guide and the skills/commands documentation at code.claude.com/docs. Behavior and version gates change frequently; verify against upstream before shipping.

---

What Looper provides

Looper provides loop design discipline: a clear goal, context sources, checkable verification, reviewer/judge gates, termination guards, a portable spec, a same-session execution handoff, and lightweight run state/log files.

Looper does not provide durable orchestration. It does not schedule cron jobs for you, persist step-level retries across process restarts, manage sub-agent lifecycles, enforce concurrency controls, or store a production run history. If you need those guarantees, use Looper to design the loop and hand the resulting spec to an orchestrator built for durable execution.

Healthy loop checklist

Before running a loop, Looper pushes you to make these explicit:

• Goal: what outcome the loop is trying to produce.
• Context: which files, commands, issues, or external sources the loop may inspect.
• Actions: which model, tools, commands, or human handoffs may change state.
• Feedback: which programmatic checks, judges, reviewers, or humans decide whether work is good enough.
• State: where the loop records status, decisions, blockers, and outputs.
• Stop conditions: success, max iterations, revision caps, no-progress signals, and budget caps. The external runner enforces wall-clock caps; token/USD caps are operator-visible advisory limits unless you add accounting around the configured CLIs.
• Execution boundary: current workspace, branch/worktree, external runner, or a separate durable orchestrator.

---

Quick start

Install as a global personal skill and slash command.

On Windows PowerShell:

irm https://raw.githubusercontent.com/ksimback/looper/main/install.ps1 | iex

On macOS/Linux:

curl -fsSL https://raw.githubusercontent.com/ksimback/looper/main/install.sh | bash

If you prefer to inspect each step, use the manual install:

Manual install commands

Windows PowerShell:

git clone https://github.com/ksimback/looper "$env:USERPROFILE\.claude\skills\looper"
New-Item -ItemType Directory -Force "$env:USERPROFILE\.claude\commands" | Out-Null
Copy-Item "$env:USERPROFILE\.claude\skills\looper\commands\looper.md" "$env:USERPROFILE\.claude\commands\looper.md" -Force

macOS/Linux:

git clone https://github.com/ksimback/looper "$HOME/.claude/skills/looper"
mkdir -p "$HOME/.claude/commands"
cp "$HOME/.claude/skills/looper/commands/looper.md" "$HOME/.claude/commands/looper.md"

Then, in Claude Code:

/looper

Looper interviews you, writes the artifacts into a folder called looper-output, and shows you an ASCII flow preview to confirm before anything is finalized. The installer also creates a private .venv inside the skill directory and installs PyYAML, which the helper compiler needs to read loop.yaml. It then offers to run the loop right there in the same Claude Code session.

If you want a different folder name, pass it after /looper, for example /looper client-onboarding-loop.

Start from a pattern template

Instead of a blank interview, start from a named, pre-designed loop:

/looper my-review --template code-review

Template	Use when
security-scan	Read-only sweep of a repo for secrets, PII, and vulnerabilities → triaged SECURITY-FINDINGS.md.
code-review	Review a branch's diff against its base → typed, severity-rated REVIEW.md grounded in the diff.
bug-hunt	Reproduce a reported bug, fix the root cause, prove it with before/after repro evidence.
docs-sync	Find and fix doc/code drift → per-item DRIFT-REPORT.md; docs follow code, code untouched.
research-synthesis	Synthesize collected sources into a cited REPORT.md; every claim traceable to a file.

Each template is a complete, compiler-validated loop.yaml with a handful of {{PLACEHOLDER}} slots; the wizard asks only for those, picks models from what's installed, and still runs its full critique, privacy, and preview flow before emitting. See templates/loops/ for the catalog and per-template docs — including how to add your own.

Lint any loop.yaml

The design rubrics also exist as a static checker — no wizard, no interview: