What is the main benefit of the Continuous Agent Loop?

It provides a structured framework for AI agents to work autonomously while ensuring quality through predefined 'gates' and recovery steps if the agent gets stuck.

How does the loop selection process work?

It automatically chooses the best workflow—such as RFC decomposition for complex plans or sequential for simple tasks—based on the specific requirements of the project.

What happens if the agent enters an infinite retry loop?

The skill includes recovery mechanisms that freeze the loop, run a harness audit, and narrow the scope to the failing unit to prevent cost drift and wasting resources.

Can I use this with existing CI/CD pipelines?

Yes, the 'continuous-pr' mode is specifically designed for environments requiring strict CI/PR controls and integration.

Continuous Agent Loop

Name: Continuous Agent Loop
Author: affaan-m

byaffaan-m

•

137,500

•

生产力与工作流

Orchestrates autonomous AI agent cycles with built-in quality gates, evaluation frameworks, and recovery mechanisms for complex development tasks.

This skill implements a standardized autonomous loop architecture designed for Claude Code, enabling the agent to handle complex, multi-step tasks with high reliability. It features a sophisticated selection process that routes tasks through specific workflows like RFC decomposition, parallel exploratory generation, or strict CI/PR pipelines. By incorporating automated quality gates and recovery protocols, it prevents common failure modes such as infinite retry loops and cost drift, ensuring that autonomous development remains productive, measurable, and safe.

主要功能

01137,500 GitHub stars

02Integrated quality gates and evaluation harnesses to ensure code integrity

03Multi-path loop selection including Sequential, PR-controlled, RFC-based, and Parallel modes

04Resource optimization to prevent infinite retries and unnecessary cost drift

05Automated recovery protocols with loop freezing and scope reduction for failing units

06Session persistence and auditing capabilities for long-running autonomous tasks

使用场景

01Managing autonomous PR creation and management within strict CI/CD environments

02Executing exploratory coding research where parallel generation of multiple solutions is required

03Automating complex refactoring tasks that require multi-stage planning and verification

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add affaan-m/everything-claude-code continuous-agent-loop

For use in Claude.ai and ChatGPT

主要功能

01137,500 GitHub stars

02Integrated quality gates and evaluation harnesses to ensure code integrity

03Multi-path loop selection including Sequential, PR-controlled, RFC-based, and Parallel modes

04Resource optimization to prevent infinite retries and unnecessary cost drift

05Automated recovery protocols with loop freezing and scope reduction for failing units

06Session persistence and auditing capabilities for long-running autonomous tasks

使用场景

01Managing autonomous PR creation and management within strict CI/CD environments

02Executing exploratory coding research where parallel generation of multiple solutions is required

03Automating complex refactoring tasks that require multi-stage planning and verification