A terminal meta-agent that watches how you work, finds the patterns, and writes the loops so you don't have to.

loopyagent.com

"I don't prompt Claude anymore. I have loops running that prompt Claude and figure out what to do. My job is to write loops."

— Boris Cherny, creator of Claude Code at Anthropic

"Stop prompting coding agents. Start designing the loops that prompt them."

— Peter Steinberger, founder of OpenClaw (formerly Warelay)

The top engineers using Claude Code and Codex aren't manually prompting back and forth. They're building autonomous loops — programs that observe, decide, and act on their behalf. The problem: most developers don't know where to start. Finding the right things to automate takes observation, pattern recognition, and time you don't have.

loopy does that work for you.

loopy is a local meta-agent that runs quietly in your terminal alongside Claude Code. It watches your sessions, spots work you keep doing by hand, and proposes ready-to-install automation loops — each one a self-contained script with its own trigger, operating instructions, and install record.

You review proposals in the dashboard. You approve the ones that make sense. The loop runs.

Everything stays on your machine. The only LLM calls go through your own claude -p binary — loopy never phones home.

Most Claude Code users are leaving 40–70% of their potential on the table by staying in prompt-response mode. Here's what shifts when you move to loops:

A single well-chosen loop replaces 50–200 manual prompts per month. At 3 seconds per prompt cycle, 3 active loops save you roughly 15–30 minutes per day — compounding as your loop library grows.

Early loopy users report discovering 3–5 automatable patterns per month they would never have spotted manually. Most of those become 10-minute installs.

loopy is a small local pipeline that runs continuously in the background:

Claude Code sessions
       │
       ▼
  [watcher] — notices new session transcripts via launchd daemon
       │
       ▼
  [digester] — compresses + redacts each session to a compact text digest
       │
       ▼
  [engine] — sends digests to your own claude -p CLI, looks for recurring patterns
       │
       ▼
  [inbox] — good candidates land as proposals; you review and approve
       │
       ▼
  [installed loop] — loop.md + trigger + manifest wired into Claude Code or Codex

The engine uses your own Claude API credits. No separate service, no subscription, no cloud component.

curl -fsSL https://raw.githubusercontent.com/secretbuilds/loopy/main/install.sh | bash

Then finish setup:

loopy setup

The installer clones the repo to ~/.loopy-app, builds it, and links the loopy binary to your PATH. Re-running the same command updates to the latest version.

Options for setup:

loopy setup --companion manual   # no automatic nudges
loopy setup --no-daemon          # configure without the background daemon

Requirements: Node ≥ 20, git, Claude Code CLI (claude) in your PATH, macOS.

loopy

Running loopy with no arguments opens the full-terminal hub:

The header shows your agent (Loopy) with live status: sessions watched · daemon state · today's token spend vs cap.

Three panels:

• inbox — pending loop proposals. Select one to see its summary, estimated impact, evidence count, and confidence score. Approve, dismiss, or snooze.
• loops — your installed loops, with trigger kind and target tool.
• activity — scrolling log of everything loopy has done in the background.

Keys:

The dashboard resizes with your terminal. It needs at least 60×16 — below that it shows a hint to grow the window.

The installer also drops a /fable slash command into ~/.claude/commands/ so it's available in any Claude Code session, globally.

/fable <your prompt>

This routes your prompt through the full Claude Fable 5 system prompt and model, inline in your current session. The response comes back as if you'd switched to Fable 5 for that one message — without leaving your session or changing your model.

/fable make this component prettier
/fable rewrite this in a more conversational tone
/fable explain this to a non-technical stakeholder

Under the hood it spawns claude -p --model claude-fable-5 --system-prompt-file ~/.loopy/prompts/fable.md as a subprocess and returns the output inline. The Fable 5 system prompt is stored at ~/.loopy/prompts/fable.md (installed automatically — no setup needed beyond running the installer).

An approved proposal becomes a bundle at ~/.loopy/loops/<id>/:

loop.md          — operating instructions fed to Claude
trigger.json     — schedule, hook, or manual trigger metadata
manifest.json    — evidence, target tool, every path installed
state/           — loop-local state (persists across runs)

The manifest is what makes loopy uninstall exact: it removes only the paths it created for that specific loop, with no guesswork.

Transcripts stay on your machine. Always.

Before any digest is sent to an LLM, loopy redacts:

• API keys, tokens, passwords, bearer tokens
• GitHub tokens, AWS keys, URL credentials
• High-entropy strings that look like secrets

The engine sends only compact redacted digests to your own claude -p process. loopy does not contact any external service.

• Claude Code and Codex power users who want to graduate from manual prompting to loop engineering but don't know where to start
• Developers shipping AI-assisted projects who want to systematically reduce the cognitive overhead of repetitive prompting
• Anyone who's read Boris's workflow post and thought: I want that, but I can't find my own patterns yet

loopy gives you the observation layer. You bring the judgment. The loops handle the rest.

loopy may suggest automation, but it never shames you for ignoring, snoozing, or dismissing a proposal. A quiet tool is better than a nagging one. Your inbox, your call.

git clone https://github.com/secretbuilds/loopy.git
cd loopy
npm install
npm run build
npm link
loopy setup

Run the full test suite:

npm run typecheck && npx vitest run

Manual smoke render (verify dashboard geometry):

npx tsx scripts/dash-smoke.ts 120 40

Live proposal-quality eval (requires claude CLI):

npx tsx scripts/live-eval.ts

loopy is early software. It watches Claude Code sessions on macOS via launchd. Windows/Linux daemon support is on the roadmap.