[2604.03515] Inside the Scaffold: A Source-Code Taxonomy of Coding Agent Architectures

Computer Science > Software Engineering arXiv:2604.03515 (cs) [Submitted on 3 Apr 2026] Title:Inside the Scaffold: A Source-Code Taxonomy of Coding Agent Architectures Authors:Benjamin Rombaut View a PDF of the paper titled Inside the Scaffold: A Source-Code Taxonomy of Coding Agent Architectures, by Benjamin Rombaut View PDF HTML (experimental) Abstract:LLM-based coding agents can localize bugs, generate patches, and run tests with diminishing human oversight, yet the scaffolding code that surrounds the language model (the control loop, tool definitions, state management, and context strategy) remains poorly understood. Existing surveys classify agents by abstract capabilities (tool use, planning, reflection) that cannot distinguish between architecturally distinct systems, and trajectory studies observe what agents do without examining the scaffold code that determines why. This paper presents a source-code-level architectural taxonomy derived from analysis of 13 open-source coding agent scaffolds at pinned commit hashes. Each agent is characterized across 12 dimensions organized into three layers: control architecture, tool and environment interface, and resource management. The analysis reveals that scaffold architectures resist discrete classification: control strategies range from fixed pipelines to Monte Carlo Tree Search, tool counts range from 0 to 37, and context compaction spans seven distinct strategies. Five loop primitives (ReAct, generate-test-repair, plan-e...