autonomous-loops

Autonomous loops in Claude Code are configured by specifying a loop pattern and execution parameters, enabling continuous integration workflows that run development tasks without human intervention. You can build sequential pipelines or complex multi-agent coordination.

Available loop patterns for code generation include sequential pipelines, REPL sessions, infinite loops, PR loops, de-sloppify patterns, and RFC-driven DAG orchestration. These span from simple sequential execution to complex multi-agent coordination.

Yes, Claude Code is required as a dependency to run autonomous development loops. The Skill designs and orchestrates loop patterns specifically for Claude Code execution, supporting automated workflows like code generation, testing, and refactoring.

Yes, multi-agent coordination is supported for parallel task execution in Claude Code. The Skill includes RFC-driven DAG orchestration patterns, allowing you to design complex autonomous workflows that coordinate multiple agents simultaneously.

The best way to automate repetitive refactoring tasks is implementing an autonomous loop pattern like the de-sloppify pattern in Claude Code. This enables continuous development workflows that perform code cleanup and refactoring autonomously.

You should avoid infinite loop patterns when tasks have a defined completion criteria or finite scope. Instead, use sequential pipelines or PR loops for bounded workflow automation, reserving infinite loops for continuous, ongoing development processes.