Skip to content
Agentic Levels
Start hereLevelsTracksPromptsAbout
Get my levelGet my level

  • Start hereYour first useful task with AI.
  • Level quiz8 questions. No AI needed.
  • Fluency ScoreYour Level and a score, from your own AI.

  • LevelsThe 11 levels, in order.
  • LessonsShort lessons, filter by level.
  • TracksGuided paths by tool, role or goal.
  • Set up your AI tool
  • Starter packFree files that help AI remember.

  • PromptsReady prompts to copy and adapt.
  • ChatGPT vs Claude vs GeminiPick the right one for the job.
  • How-tos
  • ReferenceCommands for Claude Code, Codex, Antigravity.
  • ResourcesHand-picked reading and tools.

  • Your progress
  • Your next moveAn interview that picks your next skill.
  • Your next lessonsThe next lessons at your level.
  • Updates

  • Pricing
  • About
  • FAQ & Feedback
  • Work with us
  • How we handle your data
Sign inPreferences
AboutPricingHow-tosPromptsStarter packMy stackHow we handle your data

© 2026 Fuentes Studio

PrivacyTermsRefundsAccessibilityAI content policyCredits
yourCouncil
Ready to help
✦

What do you want to understand?

Ask anything about what you're learning.

←Claude Code Fundamentals
  1. Tracks
  2. ›Claude Code Fundamentals
  3. ›How-to guide

How to use loop-engineering

A tool-agnostic reference for designing the loop that runs your AI coding agent, instead of you prompting it one turn at a time.

Levels 8 to 10On GitHub

Once your AI coding agent stops feeling new, you hit a different wall. You're still the one clicking go, every time. loop-engineering scores your setup for you, a Loop Ready score from 0 to 100, plus a checklist for turning one task you do by hand into one the agent runs on a schedule, checks itself, and reports back.

Before you start

Comfortable naming what a skill (a saved how-to Claude draws on automatically) and a sub-agent (a second Claude instance that checks the first one's work) are. New to those words? Start with Add skills and subagents with restraint first. You also need Claude Code running in a real project folder with Node installed, and loop-engineering is a third-party, MIT-licensed tool from one maintainer, not an Anthropic product.

  1. 1Get your baseline score before you change anything
  2. 2Scaffold whatever the audit says you're missing
  3. 3Estimate the cost before you schedule anything
  4. 4Start at report-only, not auto-fix
  5. 5Re-audit after you close a gap

Steps

01

Get your baseline score before you change anything

Run the audit tool against your own project. It prints a single Loop Ready score from 0 to 100, a readiness level, and a findings list naming what's missing across automations, worktrees, skills, connectors, sub-agents, and memory (the durable file that survives after the chat window closes). You can't tell what to fix until you see the number.

Ask Claude this
Run `npx @cobusgreyling/loop-audit . --suggest` in this project and tell me my Loop Ready score and what it suggests I add first.
02

Scaffold whatever the audit says you're missing

loop-init creates the starter files for whichever pattern you pick (a daily-triage pattern is the gentlest first loop) plus a budget and run-log file, so you're not hand-writing boilerplate. The command below names Claude; swap in `codex` if that's the agent you're running instead.

Ask Claude this
Run `npx @cobusgreyling/loop-init . --pattern daily-triage --tool claude` and show me what it created.
03

Estimate the cost before you schedule anything

Sub-agents and long-running loops can burn tokens fast. loop-cost estimates spend for a pattern and cadence before a single automated run happens. A surprise bill isn't how you want to find out it was expensive. (`--level L1` here is the tool's own L1-L3 readiness scale, unrelated to Agentic Levels.)

Ask Claude this
Run `npx @cobusgreyling/loop-cost --pattern daily-triage --level L1 --cadence 1d --orchestration maker-checker` (the doer-plus-checker pattern this pack teaches, not the cheaper single-agent default) and tell me roughly what this will cost per day.
04

Start at report-only, not auto-fix

The framework's own rollout is report first, then assisted fixes, then unattended. Your first loop should only read your project, write to a memory file, and tell you what it found, nothing else, for at least a week.

Ask Claude this
Set up a loop that runs once a day, updates a STATE.md file with what changed, and doesn't auto-fix anything yet.
05

Re-audit after you close a gap

Run the same audit command again after you add whatever was missing (a checker sub-agent, a scheduled run, a memory file). Watching the score move from a real 0-100 number is how you know a change actually mattered, not just that it felt like progress.

Ask Claude this
Run `npx @cobusgreyling/loop-audit . --suggest` again and tell me what moved since my last score.

How do you know it worked?

  • You have a Loop Ready score for your own project from loop-audit, not a guess.
  • You can read the audit's findings list and name what it says you're still missing.
  • You have run one loop at report-only level and it wrote to a memory file without touching anything else.

If something looks off

Skipping straight to auto-fix
Running an unattended loop before you've read a week of reports is how a small mistake compounds overnight with nobody watching. Report-only first is how you learn whether the loop is trustworthy before it acts without you.
Token costs can climb fast
Sub-agents and long-running loops multiply how much a single loop run costs. Run loop-cost before you schedule anything, not after you get the bill.
An old memory file doesn't mean you're ready for unattended
The score weighs how recently a loop actually ran heavier than whether the memory file exists at all. A `STATE.md` you wrote a month ago and never touched again isn't proof the loop still works. It just means nobody has re-checked it.
A newer single-command front door exists
This how-to uses loop-engineering's original separate commands (`loop-audit`, `loop-init`, `loop-cost`). The tool now also ships one `npx @cobusgreyling/loop <init|doctor|status|audit|cost>` front door that does the same things, if you'd rather type one command name.
No ChatGPT or Gemini equivalent yet
loop-engineering's `--tool` flag covers several agent CLIs (Claude, Codex, and others), but plugins, skills, and sub-agent dispatch are all agent-CLI features. There's no ChatGPT or Gemini equivalent to point to yet.

See also

  • Claude Code Fundamentals: skills and subagents from scratch
  • gbrain: give your agent a memory layer that survives sessions

What you'll do, 5 steps

  1. 1Get your baseline score before you change anything
  2. 2Scaffold whatever the audit says you're missing
  3. 3Estimate the cost before you schedule anything
  4. 4Start at report-only, not auto-fix
  5. 5Re-audit after you close a gap

Was this helpful?