loop-library logo

loop-library

A library of practical AI-agent loops and an installable skill for finding, adapting, and designing repeatable agent workflows.

Website GitHub

What is it?

What it is

Loop Library is a public catalog of loops (bounded, feedback‑driven prompts) plus an optional installable skill called Loopy that helps AI agents discover, audit, adapt, run, debrief, save, and publish loops using that catalog.

Why it exists

To give AI agents a repeatable, trustworthy way to improve work by providing clear success checks, stopping points, and bounded feedback cycles, reducing guesswork and enabling reuse of proven workflows.

Who should use it

Engineers, product managers, and AI developers who want to automate iterative improvement tasks, discover reusable workflows, audit and adapt existing loops, or publish new loops to a shared catalog.

Who should avoid it

Users who lack Node.js or the ability to run the Loopy skill, or who prefer manual, one-shot prompts over structured, bounded loops.

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

Discover repeated work and turn into bounded loop
Find published loops from live catalog
Audit and repair loops (Loop Doctor)
Adapt existing loops to tools and success criteria
Craft new loops via guided interview
Run loops in bounded passes with evidence receipt
Debrief run receipts for improvement
Save loops to project LOOPS.md for reuse
Prepare loops for publication with preview and approval
Browse public loop library website
Access JSON/plain-text catalog and agent guides
Works with Codex, Cursor, Claude Code agents
Requires Node.js and npx for installation

Advantages

  • Provides clear stopping criteria and feedback loops
  • Reduces guesswork in agent tasks
  • Enables reuse of proven workflows across agents
  • Supports bounded, safe automation without open-ended loops
  • Facilitates collaboration by sharing loops in library
  • Generates evidence-backed receipts for auditability
  • Allows incremental improvement via debrief
  • No installation needed to browse library
  • MIT licensed open source

Disadvantages

  • Requires Node.js and npx to install Loopy skill
  • Guided workflow only available after installing Loopy
  • Dependent on external live catalog for recommendations
  • May need agent restarts after installation
  • Learning curve to understand loop concepts
  • Limited to agents that support the skill system (Codex, Cursor, Claude Code)

Installation

FAQ

What is Loop Library and what does it contain?

Loop Library is a public catalog where people and agents can browse, read, and copy loops—playbooks with built‑in feedback. It consists of a live website (https://signals.forwardfuture.com/loop-library/) and an optional installable guide called Loopy that helps agents work with the catalog.

What is Loopy and what does it do?

Loopy is an installable skill that gives agents direct access to the Loop Library. It can discover repeated work, find published loops, audit and repair loops, adapt loops to your tools, craft new loops via interview, run loops with receipts, debrief runs, save loops locally, and prepare loops for publication—all while checking the live catalog for recommendations.

How do I install Loopy for my agent?

Install Loopy with npx skills add Forward-Future/loopy --skill loopy --agent <agent> [-g] [-y]. Examples: Codex: npx skills add Forward-Future/loopy --skill loopy --agent codex -g -y; Cursor: same with --agent cursor; Claude Code: same with --agent claude-code. List multiple --agent flags to install for several agents at once. Omit -g for project‑only install, omit -y to review prompts interactively.

How do I invoke Loopy after installation?

Invoke Loopy using your agent’s slash‑command style: In Codex type /skills, choose Loopy, then enter your request (or use $loopy). In Cursor type / in Agent chat, search for loopy, select it, and add your request (or type /loopy directly). In Claude Code type /loopy followed by your request. You can also describe a matching task normally; the agent will load Loopy when the request clearly calls for it.

What are the main paths Loopy can take when invoked?

Loopy offers nine paths: Discover (find repeated work and turn it into a loop), Find (search the live catalog for up to three published loops), Loop Doctor (audit a loop you provide and repair material problems), Adapt (tailor a published loop to your tools and success criteria), Craft (interview you to create a new loop when no good fit exists), Run (execute a loop in bounded passes and return an evidence‑backed receipt), Debrief (analyze run receipts and recommend the smallest justified improvement), Save (store a loop in your project’s LOOPS.md for later reuse), Publish (check quality, catalog overlap, prepare a preview, and submit only after your explicit approval).

Loading documentation…
View on GitHub

Featured in Videos

YouTube tutorials and walkthroughs for loop-library

Alternatives

Similar projects ranked by category, topics, and text overlap.

Compare
loop-library | MushyBook