ℹ️ About Loops

🔄 What is a Loop?

A loop is a pre-built kickoff prompt that instructs an AI coding agent (like Cursor, Claude Code, or Codex) to perform a task iteratively. The agent checks its progress after each iteration and continues until the exit condition is met — or the maximum number of iterations is reached.

Think of loops as recipes for autonomous coding. Instead of manually guiding the agent step-by-step, you give it a loop and let it self-pace through the work.

⚡ How Do Loops Work?

Every loop has the same basic structure:

• Goal — What the agent is trying to achieve
• Max iterations — How many times the agent will try before stopping
• Check command — The command the agent runs to check progress between iterations
• Exit condition — When to stop (e.g., "all tests pass")
• Steps — What the agent does each iteration

🤖 Compatible Agents

Loops work with any AI coding agent that supports multi-turn conversations. The most popular agents include:

• Cursor — AI-first code editor
• Claude Code — Anthropic's terminal-based coding agent
• Codex — OpenAI's coding agent
• Gemini CLI — Google's terminal coding assistant
• OpenCode — Open-source terminal coding tool

🛡 Hardened Loops

Some loops are marked as "Hardened". These have been tested extensively and include additional guardrails to prevent common failure modes. They are recommended for production workflows.

📎 Credits

Loops data is sourced from loops.elorm.xyz, an open community project by elorm. We sync the latest loops regularly to keep this hub up-to-date.