Guide · Agent skill
How to use the Loop Me Workflow Designer agent skill
Updated · written with AI from the project's own documentation and checked against our catalogue data
Loop Me is an agent skill that interviews you about repeated activities you might hand off to an AI agent, then writes each one down as a workflow spec file. It is for people who want to find routine work worth automating and want the agent to ask hard questions before anything gets built. After this guide you will know what the skill does, how to install it, how to trigger it, and what it does and does not change in your files.
Loop Me Workflow Designermattpocock/skills · MIT Download the skill Quick start
- 1Download and unzip. Download the loop-me ZIP from this page and unzip it so you have a loop-me folder containing SKILL.md and agents/openai.yaml.
- 2Place the skill. Move the folder into ~/.claude/skills/ for personal use, or into .claude/skills/ inside a project. Alternatively, upload it in the Claude apps under Settings → Capabilities → Skills.
- 3Set up the workspace. Make sure your workspace is writable and has a workflows folder and a NOTES.md file, even if NOTES.md is nearly empty.
- 4Start a session. Ask the agent to use loop-me, naming a workflow to design or asking it to find one.
- 5Answer the rounds. Respond to each round of questions, accepting or changing the recommended answer, until the skill finishes the spec.
- 6Review the files. Check the created or edited files in workflows/ and the terms in NOTES.md.
What the loop-me skill makes the agent do
Loop Me runs a structured interview. The agent asks one round of pointed questions at a time, and attaches a recommended answer to each one so you can accept it or change it. The interview is aimed at a repeated activity you want to delegate, and the skill treats your life or work as a set of recurring patterns (loops) that may be predictable enough to automate.
The output is spec files, not results. Each workflow is saved as a markdown file in a workflows folder, and the skill keeps a NOTES.md file with your own terms. The skill is designed to produce specs only; it does not perform the workflows it describes.
- Asks one round of questions at a time, each with a suggested answer
- Proposes routines you may not have noticed
- Creates, edits and deletes files in workflows/*.md as answers settle
- Records your terminology in NOTES.md
- Continues until each spec is complete enough for another agent to build without further questions
Install the loop-me skill from the ZIP
Download the loop-me ZIP from this page and unzip it. The skill folder contains agents/openai.yaml and SKILL.md. Place the unzipped folder in one of two locations: your personal skills folder at ~/.claude/skills/, or a project-specific .claude/skills/ folder inside the project you are working in.
If you use the Claude apps instead of a local folder, open Settings, go to Capabilities, and upload the skill under Skills.
- Personal install: ~/.claude/skills/ (available across your projects)
- Project install: .claude/skills/ inside the project folder (available only there)
- Claude apps: Settings → Capabilities → Skills, then upload the skill
Prepare your workspace before the first run
The skill needs a writable workspace that contains a workflows folder and a NOTES.md file. The source describes NOTES.md as raw notes on your world: the tools you use, the channels you process, and your own names for them. If NOTES.md is empty or thin, the skill is designed to interview you about your world before it writes any specs, so you can start with a blank or near-blank file.
Your agent runtime must load SKILL.md skills. The skill does not run scripts and does not use the network.
- A workflows/ folder (one .md spec per workflow)
- A NOTES.md file in the workspace
- An agent runtime that loads SKILL.md skills
How to trigger the skill with a natural request
The skill is marked so the model does not invoke it on its own, so you start it deliberately. The skill's argument hint says you can give it a workflow to design, or give it nothing and let it go looking for one. Describe the repeated activity in plain words, or ask the agent to find routines in your week.
Expect a long conversation. The skill is built for round-by-round questioning, so it suits a session where you have time to answer, not a quick lookup.
- Name a specific task: "Use loop-me to design a workflow for sorting my client inbox every morning."
- Ask it to search: "Use loop-me and find routines in my work week that are worth delegating."
- Answer with your own terms; the skill will record them in NOTES.md
Review what the skill writes
Because the skill creates, edits and deletes spec files in your workspace, read the changes before you rely on them. Check each workflows/*.md file and the NOTES.md glossary after a session. The skill's trigger, checkpoint and brief vocabulary is applied only when a workflow calls for it, and no AI, checkpoint or schedule is required unless the interview shows one is needed.
If you are not comfortable with an agent writing and deleting files in your workspace, use a copy of the workspace for the session.
- Open each new or changed file in workflows/
- Confirm NOTES.md has the terms you actually use
- Keep a backup or version control so deletions can be reversed
What people use it for
- Designing a spec for sorting an inbox each morning, deciding which items need a human decision
- Mapping recurring weekly admin, such as invoices or reports, and deciding which steps are predictable enough to hand off
- Writing down a repeated personal routine as a workflow spec before anyone builds an automation for it
- Building a shared glossary in NOTES.md so a team uses the same names for its tools and channels
Frequently asked questions
Is loop-me free to use?
The skill is published under the MIT licence, which is stated on this page. The licence applies to the skill itself; any agent runtime or Claude plan you use to run it has its own terms, which this page does not cover.
What do I need to run the loop-me skill?
You need an agent runtime that loads SKILL.md skills and a writable workspace containing a workflows folder and a NOTES.md file. The skill does not run scripts and does not use the network.
Where do I put the loop-me skill folder?
Put it in ~/.claude/skills/ for personal use across projects, or in .claude/skills/ inside a specific project. In the Claude apps, upload it under Settings → Capabilities → Skills.
Does loop-me run my workflows for me?
No. The skill only creates and edits spec files in workflows/. It is designed to produce specs, not to execute the workflows they describe.
Why does the agent keep asking questions?
The skill uses a relentless, round-by-round interview and treats a spec as done only when an implementer agent could build it without asking a question. If you need a quick answer, this skill is not the right tool.
How is loop-me different from just asking an agent to write a plan?
Loop-me asks one round of pointed questions at a time, attaches a recommended answer to each, proposes routines you may not have noticed, and saves results as workflow files and a glossary. A single request usually produces one document without that back-and-forth.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
Loop Me suits people and teams who want a disciplined interview that finds routine work worth delegating and records it as clear spec files. It is a poor fit if you want an agent to actually run tasks now, need a quick one-off answer, or are not comfortable with an agent writing and deleting files in your workspace.
Download the skill