Agentplay

Agentplay provides a visual session replay system for coding agents, letting software engineers and teams review and analyze agent behaviors in depth. By converting session logs already present on developers' local machines, it generates interactive maps and detailed timelines of coding agent activity. This enables users to step through what their coding assistants have done—including files touched, commands run, tools used, retries, errors, subagents engaged, and context changes—without needing to sift through raw log data.

The tool highlights key events within agent sessions—such as failures, retries, delays, context compaction, PR creation, and model changes—to help teams quickly find where and why issues or changes occurred. By supporting deep inspection, Agentplay is especially valuable for debugging, process improvement, and understanding how autonomous agents navigate codebases and tasks. Open source and with a simple local install, it requires no API key or remote account.

Agentplay helps teams ensure code quality, identify bottlenecks or non-obvious agent behaviors, and confidently iterate on their use of coding agents as part of development workflows.

Agentplay is aimed at software development teams and individual engineers who use coding agents—tools that autonomously write, edit, or manage code—to accelerate their workflows. It’s particularly useful for developers seeking better visibility into how these agents operate, where errors or inefficiencies happen, and how tasks are carried out step-by-step.

Engineering managers, tech leads, and anyone responsible for integrating AI-assisted coding into team processes will benefit from its session maps and timelines. It’s relevant to startups or established teams working on complex codebases, where trust, transparency, and optimization of autonomous development tools are crucial. Teams looking to debug, improve or validate coding agent performance will find immediate, practical value.