autonomous-loops

You orchestrate multi-agent workflows by applying looping patterns like sequential pipelines or RFC-driven DAGs to automate end-to-end tasks. This skill coordinates parallel agents using isolated context windows and persistent state across iterations.

An RFC-driven DAG is an orchestration pattern that uses a specification to direct autonomous Claude Code loops. It structures complex AI generation tasks into a directed acyclic graph, enabling parallel agent execution and isolated context windows for each stage.

You automate continuous PR loops by deploying looping patterns like the Continuous Claude PR Loop or Infinite Agentic Loop. These patterns utilize robust merge queues with eviction recovery and quality gates to automate end-to-end code generation workflows.

Yes, parallel agent coordination explicitly requires isolated context windows for each stage to maintain state. This approach ensures robust merge-queue mechanisms and cross-pattern composability when automating complex AI generation tasks.

Merge queue eviction is handled through built-in recovery mechanisms within the orchestration patterns. The skill provides robust merge queues with eviction recovery to ensure continuous workflow automation and persistent state across iterations.

The best way to build CI-like autonomous workflows is using orchestration patterns like NanoClaw REPL or De-Sloppify. These patterns provide context persistence, quality gates, and cross-pattern composability for end-to-end task automation.