Loop Library — Productivity

Loop Library helps users discover, find, audit, repair, adapt, or design repeatable AI-agent loops with explicit triggers, actions, verification, stopping conditions, guardrails, and handoffs. It analyzes codebases or coding-thread history for repeated work, recommends published loops, diagnoses weaknesses in existing loops, and runs a focused interview to craft new bounded loops. The skill produces concise loop definitions with a one-sentence explanation and a short prompt.

loop-library claude-code

📂 View skill on GitHub ↗ 🚀 How to use

✨ What it does

Routes requests to discover, find, audit, adapt, or design loops.
Analyzes codebases and thread history for repeated work, requiring at least two concrete occurrences for thread-derived
Searches the live Loop Library catalog for published loops and ranks matches by outcome, inputs, verification, authority
Audits loops against feedback-cycle rules: observable success gates, explicit terminal states, no-progress stops, approv
Runs a plain-language design interview to gather only essential details and crafts a minimal loop prompt.
Validates every crafted loop by silently tracing one cycle and repairing material gaps.

🎯 When to use it

When a user wants to analyze a codebase or coding-thread history for repeated work that could become an automated loop.
When a user asks to find or recommend a published loop for a specific outcome or problem.
When a user wants to audit or repair an existing loop for weak checks, unsafe authority, unbounded repetition, stale state, or unclear stopp
When a user wants to create a recurring agent workflow or automation cadence from scratch or by adapting a published loop.
When a user wants to turn a repeated engineering task into a bounded, copy-ready loop.

🚀 How to use

Trigger the skill by asking to discover, find, audit, repair, adapt, or design a loop. Provide the codebase, thread history, or loop definition to analyze, or describe the outcome you want. The skill may ask one short question at a time if details are missing. Example prompts:

Analyze my repository for repeated tasks that could become automated loops.
Find a published loop for generating weekly status reports.
Audit this loop for weak stopping conditions: [paste loop].
Design a loop that reviews pull requests and stops when no issues remain.

📄 Output: A concise loop definition in markdown with a one-sentence explanation and a short prompt, or a list of recommended published loops with links and adaptations.

📦 Add this skill to Claude Code

# 1. Get the skills repo
git clone --depth 1 https://github.com/alirezarezvani/claude-skills /tmp/claude-skills

# 2. Copy this skill into your project (or ~/.claude/skills for all projects)
mkdir -p .claude/skills
cp -r /tmp/claude-skills/loop-library .claude/skills/loop-library

Skill source: loop-library/SKILL.md

⚠️ Good to know

Requires access to the live catalog for published-loop discovery; otherwise it reports unavailability and does not substitute repository content. It never invents loop titles, URLs, or details and stays grounded in user-supplied or scoped evidence.

❓ FAQ

What does the skill do if the live catalog is unavailable?

It says that published-loop discovery is temporarily unavailable and does not use repository content or memory as a substitute for the production database.

How does the skill decide if work is repeated enough to become a loop?

For thread-derived candidates, it requires at least two concrete occurrences of semantically equivalent work before calling it repeated.

Can the skill activate a loop or schedule it?

No, designing a loop does not authorize enabling a schedule, changing production, or sending external messages; implementation happens only when the user explicitly asks.

🤖 Overview, features, install steps and FAQ were generated from the project's SKILL.md on Sep 4, 2026. Always check the original source before running commands.