Interactive Debugging and Steering of Multi-Agent AI Systems

Will Epperson, Gagan Bansal, Victor Dibia, Adam Fourney, Jack Gerrits +2 more
3/3/2025
cs.MAcs.AIcs.HC

No official code found. Best implementation: microsoft/agdebugger (73 stars, MIT, last commit 2026-02-11).

Abstract

Fully autonomous teams of LLM-powered AI agents are emerging that collaborate to perform complex tasks for users. What challenges do developers face when trying to build and debug these AI agent teams? In formative interviews with five AI agent developers, we identify core challenges: difficulty reviewing long agent conversations to localize errors, lack of support in current tools for interactive debugging, and the need for tool support to iterate on agent configuration. Based on these needs, we developed an interactive multi-agent debugging tool, AGDebugger, with a UI for browsing and sending messages, the ability to edit and reset prior agent messages, and an overview visualization for navigating complex message histories. In a two-part user study with 14 participants, we identify common user strategies for steering agents and highlight the importance of interactive message resets for debugging. Our studies deepen understanding of interfaces for debugging increasingly important agentic workflows.

View on arXivView PDF

Code Implementations(1)

microsoft/agdebuggerCommunity100%
MIT
736CSS, HTMLJan 29, 20258 months agoMIT

Similar Papers

DIG to Heal: Scaling General-purpose Agent Collaboration via Explainable Dynamic Decision Paths

80% match2026

Understanding Software Engineering Agents: A Study of Thought-Action-Result Trajectories

80% match1 repo152025

Exploring Autonomous Agents: A Closer Look at Why They Fail When Completing Tasks

80% match1 repo72025

Debug2Fix: Supercharging Coding Agents with Interactive Debugging Capabilities

80% match2026

Your Co-Workers Matter: Evaluating Collaborative Capabilities of Language Models in Blocks World

80% match1 repo22024

AgentMesh: A Cooperative Multi-Agent Generative AI Framework for Software Development Automation

80% match2025

Cite this paper

@article{epperson2025interactive,
  title  = {Interactive Debugging and Steering of Multi-Agent AI Systems},
  author = {Will Epperson and Gagan Bansal and Victor Dibia and Adam Fourney and Jack Gerrits and Erkang Zhu and Saleema Amershi},
  year   = {2025},
  eprint = {2503.02068},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2503.02068v1}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2503.02068.svg)](https://codeofpaper.com/paper/2503.02068)

Discussion