Agent ATO visualizes AI coding agent actions from logs
Agent ATO: Visualizing Agent Interaction Timelines from Logs
Software Engineering
Summary
AI coding helpers do many actions during coding tasks, but just seeing their final code changes doesn't show what they did step by step. The authors created Agent ATO, a tool that reads the AI's activity logs and turns them into easy-to-understand timelines. These timelines show when the AI searched, read, edited, or ran code, helping developers see the AI’s thinking process. This makes it clearer how the AI works and how different runs compare.
What this means in practice
- •For software developers: Inspect AI coding agents’ step-by-step actions during automated repair tasks to better understand and debug agent behavior.
- •For devops engineers: Compare execution patterns of AI agents running tests and builds from logs to optimize continuous integration workflows.
Authors
Takuto Kawamoto, Yoshiki Hig, Raula Gaikovina Kula
Abstract
AI coding agents are becoming part of developers' workflows, but their behavior is difficult to understand from final code changes alone. During a task, agents interact with software repositories through sequences of actions such as searching for files, reading code, editing programs, and running tests or build commands. These interactions, together with token usage, are often recorded in console logs, but raw logs are difficult for developers to inspect. In this paper, we propose Agent ATO (Agentic Trajectory Observer), a tool for visualizing AI coding agent interaction timelines from console logs. Agent ATO extracts agent interactions, classifies them by command or tool type, and visualizes them as timelines. In addition to an all-interaction timeline, Agent ATO provides filtered timelines that emphasize file discovery, file reading, file editing, and execution while preserving surrounding context. We illustrate how Agent ATO may help developers inspect and compare agent actions using selected runs from two repair tasks. Future work will apply Agent ATO to more agents, tasks, and development environments, and will evaluate whether it reduces the effort needed to compare trajectories.