Hedwig: Dynamic Autonomy for Coding Agents Under Local Oversight
Tanjal Shukla (University of Washington), Kevin Feng (University of Washington), Leijie Wang (University of Washington), Mohammad Rostami (Amazon GenAI Innovation Center), Amy Zhang (University of Washington)
Security & Privacy Architectural Patterns & Composition
Summary
A CLI coding agent that dynamically adapts its autonomy level based on developer-agent interaction history, tightening oversight in unfamiliar territory and loosening it where trust is earned.
Description
Even as coding agents can handle increasingly complex tasks, their continued tendency to introduce unintended edits, subtle bugs, and scope drift that slip past code review means developers must still decide how much autonomy to grant them. However, existing approaches for setting an agent's level of autonomy, such as static permission settings or instruction files, cannot account for how developers' preferences for agent autonomy can shift across tasks and over time. We conducted a formative survey with 21 software engineers who use coding agents, finding that they experience frustration with calibrating autonomy and have evolving preferences for level of oversight. Building on these insights, we present Hedwig, a CLI coding agent that adapts its autonomy level based on developer-agent interactions across sessions. Rather than operating on a global, fixed autonomy configuration, Hedwig learns an evolving set of behavioral guidelines from developer decisions and feedback, reducing friction on work for which the agent has earned trust, while tightening oversight when the agent operates outside familiar territory. Hedwig demonstrates the potential of a new paradigm where agents intelligently adapt their autonomous behaviors based on user trust through active, longitudinal collaboration.
