The Work Behind Delegation: A Framework for Supervising AI Coding Agents
On the map: LLM agents and tool use → Software engineering coding agents
Abstract (arXiv)
As AI coding agents carry out development tasks with greater autonomy, developers are shifting from direct implementation toward supervising delegated work. Yet existing research offers limited understanding of how developers organize supervisory activities into connected workflows. Drawing on observations and workflow diagrams from 19 experienced developers, we reconfigure Sheridan's framework of human supervisory control into seven stages and the connecting loops for supervising AI coding agents. We applied the framework to public developer discussions on Reddit and found that supervisory demands extend across stages and that developers manage them by concentrating effort in planning, delegating supervisory work to other agents, and turning recurring guidance into reusable assets. Our framework provides a useful analytical lens for understanding how developers supervise AI coding agents by capturing how supervision is structured in agentic software development.
Key facts. No citing papers in the map yet (data through 7 Oct 2026). Placed in Software engineering coding agents, one of 16 directions in LLM agents and tool use. Builds on 21 of its 85 references that are in the map. First posted on arXiv 21 Sep 2026.
Builds on
21 of 85 references.
Closest ancestors · 6
Same direction, newest first
- 2026SWE-chat: Coding Agent Interactions From Real Users in the Wild22 Apr
- 2026Rethinking Software Engineering for Agentic AI Systems12 Apr
- 2026AgentStepper: Interactive Debugging of Software Development Agents6 Feb
- 2025A Comprehensive Survey on Benchmarks and Solutions in Software Engineering of LLM-Empowered Agentic System10 Oct
- 2025A Survey on Code Generation with LLM-based Agents31 Jul
- 2025Vibe Coding vs. Agentic Coding: Fundamentals and Practical Implications of Agentic AI26 May
Other references · 15
- 2025Measuring the Impact of Early-2025 AI on Experienced Open-Source Developer Productivitycited by 93
- 2026Natural-Language Agent Harnessescited by 54
- 2023Is AI the better programming partner? Human-Human Pair Programming vs. Human-AI pAIr Programmingcited by 20
- 2025Professional Software Developers Don't Vibe, They Control: AI Agent Use for Coding in 2025cited by 17
- 2026Programming by Chat: A Large-Scale Behavioral Analysis of 11,579 Real-World AI-Assisted IDE Sessionscited by 12
- 2025Good Vibrations? A Qualitative Study of Co-Creation, Communication, Flow, and Trust in Vibe Codingcited by 10
- 2025Measuring and mitigating overreliance to build human-compatible AIcited by 9
- 2026Skill Coverage: A Test Adequacy Metric for Agent Skillscited by 7
- 2026Comparing Human Oversight Strategies for Computer-Use Agentscited by 6
- 2025Interaction, Process, Infrastructure: A Unified Framework for Human-Agent Collaborationcited by 4
- 2026Evolving with AI: A Longitudinal Analysis of Developer Logscited by 4
- 2026AI Agents Push Humans Out of the Loopcited by 3
3 more
Built on by
Not cited yet.
Similar, not cited
Close in wording, no citation either way
- 2025Code with Me or for Me? How Increasing AI Automation Transforms Developer Workflows
- 2026Humans are Missing from AI Coding Agent Research
- 2026Human oversight of agentic systems in practice: Examining the oversight work, challenges, and heuristics of developers using software agents
- 2026Software Delegation Contracts: Measuring Reviewability in AI Coding-Agent Work
- 2026Hedwig: Dynamic Autonomy for Coding Agents Under Local Oversight
- 2026Are We All Using Agents the Same Way? An Empirical Study of Core and Peripheral Developers Use of Coding Agents
- 2026How Do AI Coding Agents Contribute to Software Development? an Empirical Study of Agentic Pull Requests
- 2026How Coding Agents Fail Their Users: A Large-Scale Analysis of Developer-Agent Misalignment in 20,574 Real-World Sessions