loopy
A library of practical AI-agent loops and an installable skill for finding, adapting, and designing repeatable agent workflows.
What is it?
What it is
Loop Library is a repository that provides a public catalog of loops (prompt-based workflows) and an installable skill called Loopy that helps AI agents discover, audit, adapt, run, and publish those loops.
Why it exists
To give AI agents repeatable, bounded workflows with built‑in feedback and clear stopping points, making assisted work more trustworthy, repeatable, and reusable without open‑ended guessing.
Who should use it
Developers, AI agent users, and automation engineers who want to create, discover, and refine repeatable workflows (loops) for AI agents to improve code, documentation, testing, or other repetitive tasks.
Who should avoid it
Users who only need one-off prompts, have no repetitive work to automate, or lack Node.js/npx environment and prefer not to install agent skills.
How it works
A quick walkthrough in plain English
How loopy works in 4 steps
Step 1 of 4
Something triggers the flow
A schedule, webhook, or manual click tells loopy to start.
Features
Advantages
- Provides clear stopping criteria and bounded execution for AI agent tasks
- Reduces guesswork by turning open-ended goals into repeatable loops
- Enables reuse of proven workflows across agents and projects
- Helps identify and eliminate repeated work in codebases and coding threads
- Safe by default – no automatic scheduling, deployment, or side effects
- Works with live catalog ensuring up-to-date loop recommendations
- Project‑local saving via LOOPS.md keeps custom loops private
- Guided interview (Craft) creates loops without prior loop knowledge
- Audit and repair features improve loop quality and safety
- Multi‑agent support via skill system
- No installation of the Loop Library website required
- Transparent evidence‑backed receipts after each run
Disadvantages
- Requires Node.js and npx to install
- Dependent on external live catalog; offline use limited
- Discovery only works on repositories and coding threads the agent can access
- May need explicit invocation for predictable results
- Limited to agents that support the skill system (Codex, Cursor, Claude Code)
- Initial learning curve to understand loop concepts
- Saving loops does not grant execution permissions; separate approval needed
- Published loop submission requires manual approval step
- May not handle highly complex, non‑iterative workflows
Installation
FAQ
What is the Loop Library and how does it differ from the Loopy skill?
The Loop Library is the public catalog of loops accessible at https://signals.forwardfuture.com/loop-library/; it can be browsed without any installation. Loopy is an optional installable skill that gives an AI agent guided workflows to discover, find, audit, adapt, craft, run, debrief, save, or publish loops using that live catalog.
How do I install Loopy for my agent?
You need Node.js and npx. Run a command like `npx skills add Forward-Future/loopy --skill loopy --agent codex -g -y` for Codex, replace `codex` with `cursor` or `claude-code` for other agents, or list multiple agents in one command. Omit `-g` for a project‑local install, and drop `-y` to review prompts interactively.
How do I invoke Loopy after it’s installed?
Invocation varies by platform: in Codex type `/skills` then choose Loopy or use `$loopy`; in Cursor type `/` in Agent chat, search for loopy, or type `/loopy` directly; in Claude Code simply type `/loopy` followed by your request. You can also phrase a request naturally and the agent may load Loopy automatically if it detects a matching intent.
What are the nine main paths Loopy can take when I invoke it?
Loopy supports Discover (find repeated work and turn it into a loop), Find (search the live catalog for up to three published loops), Loop Doctor (audit and repair a loop), Adapt (tailor a loop to your tools and success criteria), Craft (interview‑driven loop creation), Run (execute a loop in bounded passes and return a receipt), Debrief (analyze run receipts for improvements), Save (store a loop in the project’s LOOPS.md), and Publish (prepare and submit a loop to the Library after approval).
How does Loopy save loops for later reuse and what is LOOPS.md?
When you ask Loopy to save a loop, it appends the accepted loop to a file named `LOOPS.md` at the project root, including a name, one‑sentence explanation, the exact prompt, and the save date. If the loop is adapted from a published Library loop, the source URL and the source’s modified date at save time are also recorded. Saved loops are project‑local, readable by Loopy in future find or craft requests, and are only created or modified on your explicit request.
How does the Discover feature work and what does it require?
Discover inspects an authorized codebase, coding‑thread history, or both for recurring work. It needs at least two distinct thread occurrences to consider work repeated, then checks whether the work has clear feedback, success criteria, limits, stopping behavior, and approval boundaries. Loopy also consults the live catalog to avoid duplicating an existing loop and returns either a new loop, an adaptation, a short candidate slate, or a clean no‑op if nothing truly fits.
Featured in Videos
YouTube tutorials and walkthroughs for loopy
Alternatives
Similar projects ranked by category, topics, and text overlap.
