Explaining Four Agent Loop Structures
theomitsa · x · 2026-07-14
This is a deep dive into agent loop engineering, arguing that while people often treat it as a single concept, it's actually about choosing different loop structures tailored to specific tasks.
The article introduces a decision framework: any loop must answer two questions—who triggers the run, and who decides when the work is done. In manual runs, humans answer both; the goal of loop design is to hand over more of these decisions to the system.
It outlines four structures:
- Turn-based: Triggered by user input, ideal for evolving requirements where each output shapes the next prompt.
- Goal-based: Triggered by goal commands like /goal, continuously pushing towards success criteria.
- Event-based: Triggered by external events, better suited for monitoring, alerts, and response tasks.
- Autonomous: The system drives itself to completion, fitting work with clear boundaries that can automatically converge.
A quote from Boris Cherny (builder of Claude Code) notes that he no longer "prompts Claude" but instead "engineers loops," highlighting a shift from writing prompts to designing agentic systems.
Related event: Deep Dive into Four Core Agent Loop Architectures(3 posts)→
More from coding & agent
- Dev builds interactive 3D product experience with GPT-6 Astra + Hyper3D Rodin — nikola_mr64990 · 2026-09-11
- Codex tip: use Sol with Astra and Luna sub-agents to save usage — pvncher · 2026-09-11
- agents-best-practices: a provider-neutral Agent Skill for designing and auditing agentic harnesses — tom_doerr · 2026-09-11
- Cognition's SWE-2 uses a KKT duality argument in RL to shift the effort Pareto curve — YouJiacheng · 2026-09-11
- First-ever Three.js Conference lands in Paris, with a panel on AI-shortened design workflows — OdinLovis · 2026-09-11
- Data engineering, not agent frameworks, is the real bottleneck for enterprise AI agents — dhruv2038 · 2026-09-11