Continuous Agent Loop

This is the canonical v1.8+ framework for running autonomous agent loops that don't spin their wheels forever. It gives you a decision tree for picking the right loop style (sequential, PR-gated, RFC-based, or exploratory parallel), then layers in quality gates, eval harnesses, and recovery patterns. The key insight here is that autonomous loops fail predictably: they churn without progress, retry the same broken thing, or rack up costs through unbounded escalation. The recovery protocol is practical: freeze the loop, audit with harness tools, reduce scope to the failing unit, and replay with explicit acceptance criteria. If you're building agents that need to run unsupervised for more than a few iterations, this gives you the control plane to keep them from going off the rails.

Install to Claude Code

npx -y skills add affaan-m/everything-claude-code --skill continuous-agent-loop --agent claude-code