What problem does it solve? Running autonomous AI agents in continuous loops often leads to churn without progress, repeated failures on the same root cause, stalled merge queues, and uncontrolled cost escalation. This Skill provides structured patterns for selecting, combining, and recovering agent loops so long-running automation stays measurable and controllable. ## Core Features & Use Cases - Loop Selection Flow: Routes work to the right loop pattern based on need: continuous-pr for strict CI/PR control, rfc-dag for RFC decomposition, infinite for exploratory parallel generation, or sequential as the default. - Combined Production Stack: Recommends a layered pipeline of RFC decomposition, quality gates, an eval harness, and session persistence for production-grade autonomy. - Failure Recovery: Defines concrete recovery steps such as freezing the loop, running a harness audit, reducing scope to the failing unit, and replaying with explicit acceptance criteria. - Use Case: A team running an overnight coding agent notices it retrying the same failing test for hours. They apply the recovery pattern: freeze the loop, audit the harness, narrow scope to the failing unit, and replay with clear acceptance criteria. ## Quick Start Ask the agent to set up a continuous agent loop with quality gates and an eval harness for your current coding task.
